Spatial continuities and discontinuities in two successive ... · 2 The RWA model of innovation and...

DEMOGRAPHIC RESEARCH VOLUME 28, ARTICLE 4, PAGES 77-136 PUBLISHED 11 JANUARY 2013 http://www.demographic-research.org/Volumes/Vol28/4/ DOI: 10.4054/DemRes.2013.28.4 Research Article

Spatial continuities and discontinuities in two successive demographic transitions: Spain and Belgium, 1880-2010

Ron Lesthaeghe

Antonio Lopez-Gay © 2013 Lesthaeghe & Lopez-Gay. This open-access work is published under the terms of the Creative Commons Attribution NonCommercial License 2.0 Germany, which permits use, reproduction & distribution in any medium for non-commercial purposes, provided the original author(s) and source are given credit. See http:// creativecommons.org/licenses/by-nc/2.0/de/

Table of Contents

1 Background 78 2 The RWA model of innovation and diffusion of new behavioral

forms

79 3 From the FDT to the SDT: The Belgian example revisited 84 3.1 The three secularization waves in Belgium: 1750-1970 85 3.2 Spatial continuity: Indicators and covariates for the Belgian

example

92 3.3 Statistical results for the Belgian example 94 3.3.1 Factor 1: Successive demographic innovations and long term

spatial continuity

95 3.3.2 Factor 2: Historical illegitimacy in fertility and the urban-industrial

connection

97 3.3.3 Factor 3: SDT fertility postponement 99 3.4 The Belgian experience in a nutshell 101 4 From FDT to SDT: The Spanish experience 102 4.1 The Spanish spatial secularization pattern 103 4.2 The demographic dimensions of the FDT and the SDT in Spain 109 4.3 Statistical results for the Spanish example 110 4.3.1 The FDT marital fertility transition 110 4.3.2 The FDT nuptiality system 113 4.3.3 The “illegitimacy” dimension of the FDT 115 4.3.4 The postponement dimension of the SDT 116 4.3.5 The “non-conformism” dimension of the SDT 119 5 A common multivariate analysis strategy for both countries 120 5.1 The principal components of the independent variables sets in

Belgium and Spain

121 5.2 Conclusions: Back to the RWA paradigm 126 6 Acknowledgements 128 References 129 Appendices 133

Demographic Research: Volume 28, Article 4

Research Article

http://www.demographic-research.org 77

Spatial continuities and discontinuities in two successive

demographic transitions: Spain and Belgium, 1880-2010

Ron Lesthaeghe1

Antonio Lopez-Gay2

Abstract

This is a study of how the synergisms between cultural and structural factors, which

played a major role during the historical fertility and nuptiality transition (first

demographic transition or FDT), have continued to condition demographic innovations

connected to the “second demographic transition” (SDT). The continuity or

discontinuity from the “first” to the “second” demographic transition is studied for

spatial aggregates over more than a century in two national contexts, i.e. Belgium and

Spain. Special attention is paid to the role of successive secularization waves in shaping

the geographical patterns of both transitions. The study also shows that the maps of the

two aspects of the SDT, i.e. the “postponement” and the “non-conformism” transitions

respectively, are shaped by different determinants. Explanations are offered using the

“Ready, Willing, and Able” paradigm, which allows us to uncover the different

conditioning and limiting factors involved. The “non-conformist” transitions (control of

marital fertility during the FDT and rise of cohabitation and non-conventional family

formation during the SDT) more closely mirror the history of secularization and the

“Willingness” condition, whereas the fertility postponement aspect of the SDT mainly

reflects female education and employment, or the “Readiness” condition. This

generalization holds in both countries. However, in Belgium spatial continuity from

FDT to SDT is connected to stable patterns of secularization, whereas in Spain it is

linked to long standing differences with respect to female literacy and education.

1 Emeritus professor Vrije Universiteit (VUB), and Royal Academy of Sciences, Brussels. Support for his work in Barcelona came from AGUAR, Generalitat de Catalunya. E-mail: [email protected]. 2 Centre d’Estudis Demogràfics, Edifici E2, 08193 Bellaterra, Spain.

Lesthaeghe & Lopez-Gay: Spatial continuities and discontinuities in two successive demographic transitions

78 http://www.demographic-research.org

1. Background

Several books and articles belonging to the so-called “Princeton Fertility Project”

(especially: Livi-Bacci 1971, 1977, Lesthaeghe 1977, Lesthaeghe and Wilson 1986) are

often cited as “cultural” corrections or complements to the classic economic and

structural interpretations that dominate studies of historical fertility decline. One of the

sources of inspiration (if not the main one) of the Princeton Fertility Project was

without doubt William Leasure’s doctoral dissertation (1962) on the Spanish fertility

transition. In this thesis Leasure showed that the classic indicators of urbanization and

industrialization largely failed to account for the regional diversity in Spain’s fertility

decline, and that regional and linguistic particularisms prevailed. This convinced A. J.

Coale that it would be worthwhile to extend this Spanish investigation to the rest of

Europe. The “Princeton Project” gradually emerged as a result, not as a cohesive project

with preconceived objectives, but as a mere collection of empirical contributions from a

variety of authors. Most of the research belonging to the Princeton project was

performed during the late 1960s and through the 1970s, and several other and similar

investigations were conducted during the following decades, including major further

major studies on Spain and Prussia (e.g., Reher and Iriso-Napal 1989, Galloway et al.

1994, Galloway 2009).

Roughly 30 years after the completion of the Princeton studies, new demographic

features have emerged with respect to household formation and fertility, and these are

by now often collectively referred to as the “Second demographic transition” (SDT)

(Lesthaeghe and van de Kaa 1986). The SDT has two main dimensions: (i) the

“postponement transition”, referring to the upward shifts in ages at marriage and at first

and subsequent births, and (ii) the “non-conformism transition”, pointing at the growth

of unconventional forms of household formation (rise of cohabitation, of fertility among

cohabitants, of single parent households, of same sex households, and of reconstituted

families based on cohabitation).

The succession of two waves of demographic innovation, referred to here as the

first and second demographic transitions (FDT and SDT), begs the question of long-

term spatial continuity (e.g., Lesthaeghe and Neels 2002). More specifically, the

question is addressed as to whether or not the leading regions during the FDT

innovation would also be the leading ones at the onset of the SDT. Spatial continuity

from FDT to SDT would be testimony as to the persistence of such long-term regional

subcultures, whereas discontinuity would be pointing at new and recent developments

in the structural and cultural determinants of the SDT3.

3 It should be stressed that at no point do we extrapolate the results of our correlation analysis from the aggregate to the individual level. Hence, the “automatic” label of “ecological fallacy”, often mindlessly

attached to all spatial analyses, would be misguided in this instance. Furthermore, a spatial analysis is, of



In order to study the nature of cultural-structural synergisms over time and space,

use will be made of A. J. Coale’s set (1973) of preconditions for innovation, otherwise

known as the “Ready, Willing, and Able” trio (RWA). The empirical data will come

from two countries: Belgium and Spain. Both countries are characterized by a high

degree of both socio-economic and cultural (linguistic and political) heterogeneity, and

hence they are ideal settings for studying regional (dis)continuities in demographic

developments.

Furthermore, we shall pay special attention to a reconstruction of the spatial

aspects of secularization in both countries, as these are intimately linked to the

willingness condition in the RWA framework.

To sum up, the three objectives are:

(i) to contribute to studies of historical continuity of demographic patterns;

(ii) to obtain historical insights into the “genesis” of the SDT;

(iii) to use the Boolean nature of the RWA paradigm to explain both continuities in

some features and discontinuities with respect to others.

2. The RWA model of innovation and diffusion of new behavioral

forms

At the end of the Princeton European Fertility Project that studied the historical fertility

and nuptiality transitions, A.J. Coale (1973) came up with a succinct and catchy

formulation of the three preconditions for a demographic transition to occur. This

clearly superseded the more detailed narratives offered by the Princeton project’s

various country studies, but caught the gist of their findings. Firstly, according to Coale,

any new form of behavior must yield benefits that outweigh the costs or disadvantages.

If there is no such economic advantage (= “Readiness” or R), then that new form of

behavior will not be attractive and there will not be a breakthrough. Secondly, the new

form must be “legitimate”, i.e. it must be culturally and ethically acceptable. If the new

form of behavior runs counter to prevailing beliefs or to religious or moral rules, then

the condition of “Willingness” (=W) will not be met. Thirdly, there must be adequate

means (e.g., of a technical or legal nature) to implement the new form. This is the

course, not the only way to look at historical change, but, in our opinion, still a salient one. Firstly, regions

have long histories and traditions or subcultures that developed over centuries, and as such they transcend the

life span of individuals. Secondly, if firms or institutions can be compared, so can other aggregates such as

geographical ones, provided that the results are interpreted at that level. Thirdly, it is not because some

theoretical paradigms, such as rational choice theory in sociology or neo-classic micro-economics, have shifted the attention to the individual-level decision making process, that all social science research should

become “atomistic” too, and thereby lose out dramatically on historical richness.



“Ability” condition (=A). The three preconditions must be met jointly for a success S

(i.e. a breakthrough of a new behavioral form) to occur:

S = R AND W AND A or S=RWA

where AND is the logical “and” or conjunction. Any failure of satisfying merely one

of the three conditions results in an overall failure, i.e., there will be no adoption or

breakthrough or transition to that new behavioral form. In other words, there are three

necessary conditions, but a single one is not sufficient. The model is not one with

separate, additional effects, but a Boolean one focusing on synergistic combinations

instead (cf. Ragin 1987). The RWA model can be specified at the micro level as well,

and this will bring out a few extra features dealing with the dynamics of innovation

processes in general and of demographic transitions in particular (Lesthaeghe

and Vanderhoeft 1999). These will be explained now in greater detail.

We assume that any individual or household i would have its own set of three

scores for, respectively, Ri, Wi, and Ai. These scores range in intensity from zero to

unity. Zero then means: no perceived advantage at all (R=0), not acceptable on moral,

religious, or other cultural grounds (W=0), and no means of implementation (A=0).

Unity corresponds to: numerous advantages completely outweigh any disadvantages

(R=1), perfectly morally and culturally acceptable (W=1), and no technical or legal

barriers (A=1). A score of 0.5 corresponds to a point of indecision. The condition for

a success is satisfied when all three scores move beyond that mid-way point, and are

hence larger than 0.5. Another way of stating this is that each individual or household

has a minimum score MINi which is the smallest of the three component scores Ri, Wi,

or Ai:

MINi= Minimum(Ri,Wi,Ai)

Hence, any actor will only adopt a new form of behavior if his MINi > 0.5. The

collection of individual scores obviously forms three distributions, respectively, for R,

W, and A, but the collection of individual minima will add a fourth distribution. This

MIN distribution will of course depend on the location and shapes of the R, W, and A

distributions, but its mean will always be lower than that of the other three. The

example in Figure 1 clarifies this point (cf. Lesthaeghe and Vanderhoeft 1999).



Figure 1: RWA model – examples of the location of the Ri, Wi, and Ai

distributions together with that of the distribution of their minima

(MINi)

Note: Location of W (left), R (middle), and A (right) at one point in time (second example) and location of the distribution of the

minimum (Ri, W i, Ai) (= dotted line).

Note: Location of W (left), R (middle), and A (right) at one point in time (third example) and location of the distribution of the minimum

(Ri, W i, Ai) (=dotted line).



In the first example, several cases have passed the 0.5 score on R or W or A, but

nobody satisfies the RWA configuration where all three conditions are being met

simultaneously. This can be seen from the distribution of minima, which does not have

an upper tail that has moved beyond the 0.5 value. In the second example most actors

know about proper ways of implementing the new form of behavior so that the

distribution of A has already shifted to the right on the 0-1 intensity scale. With

respect to readiness, the modal category is undecided (score 0.5) with half the

population still not seeing a decisive advantage. But the majority in this example

considers the new form of behavior as ethically or culturally unacceptable. The

distribution of the scores that are the smallest of the previous three is located quite a

bit further to the left than any of the other three distributions and only a small fraction

has crossed the 0.5 point. Hence, few people have made a transition to a new form of

behavior. In this example non-willingness obviously contributes disproportionately

to the lower minima, and is therefore a dominant bottleneck factor or inhibitor.

During a process of change, all four frequency distributions move from the low

end to the high end of the intensity scale on the horizontal axis, with the distribution

of the minima always trailing behind. The R, W, and A distributions can follow their

own pace, and as they shift their variances will also tend to expand. At the outset

variances are low since the vast majority has low scores on all distributions, and at the

end of the transition variances will again diminish as more and more persons end up

with high scores for every precondition. Mid-way, variances are the highest, and the

same holds for the distribution of minima. Moreover, it is likely that at that time the

MIN distribution also comes close to adopting a bell-shaped curve. If this occurs,

then our RWA model will produce a growth curve of adopters of new behavioral

forms that closely resembles Verhulst’s logistic curve (elongated S-curve). Many

innovations and their diffusion, from gothic cathedrals to engines, from epidemics to

rumors, follow such a logistic growth curve. Furthermore, the logistic curve for an

older innovation tends to taper off and reach a saturation-level of no further expansion

when new and better technologies or innovations start growing and replacing it. Also,

the latest innovation can entirely wipe out the older pattern, and in this instance there

is a new transition. And, if such transitions succeed each other, there is no problem

with numbering them as a simple means of identification.

So far we have treated the shifts in the distributions of R, W, and A as

independent. This is not likely to hold. Structural or materialist determinists, for

instance, would commonly argue that R is the leading condition and that W and A are

mere lagging derivatives. In this view, material conditions change first and people

adjust their behavior to such new circumstances and opportunities. Subsequently both

morality and technology will come under increasing pressure to adapt as well. There

are, of course, numerous examples where other sequences hold. Breakthroughs in



contraceptive and reproductive technology, for instance, have opened up avenues for

new interventions, and in this case A is the leading condition. Similarly, some cultures

may have no objections to many forms of contraception and commonly accept

abortion, and then the W-distribution will not be trailing behind the other two. To sum

up, the sub-model with R being the leading condition and with cultural and/or

technological lags may be frequently encountered, but it is by no means the only

possibility.

For each of the three diffusion patterns with respect to R, W, and A, we should

expect there to be at least one focus of initial innovation from which the diffusion

occurs until it meets social barriers. These barriers can be social class distinctions,

cultural obstacles (e.g., religious differences), or communication barriers (e.g.,

linguistic borders). From that point on, socio-economic, cultural, and spatial variables

observable at the macro level (e.g., for spatial units) will emerge as determinants of

the process of differential spa t ia l diffusion (cf. Bocquet-Appel and Jakobi 1996).

To sum up: at present we have a model of innovation based on Coale’s initial

model of three preconditions and capable of producing a logistic growth curve for any

new form of behavior. Each of the three preconditions can be “individualized” and

translated back to the macro-level in the form of shifting distributions. Moreover,

these shifts and especially differences in the pace of the shifts can be linked to

mechanisms of social and spatial diffusion of the “contagion” type, in which network

contacts are essential. Then social group and/or geographical patterns emerge in

which innovating groups or regions lead the way, and in which others follow

depending on the strength of various types of barriers. Such barriers can exist with

respect to any of the three preconditions, but since the MIN-distribution in the RWA-

model is the crucial one, it suffices for only one of the three preconditions to be

obstructed for the diffusion of the new behavioral form to be stopped or delayed at

such a social or spatial barrier. This has important consequences:

1. Those in the vanguard of a transition must score high on all three conditions

and this will set them apart from the others, for whom at least one condition is not

being met.

2. Conversely, if one of the three distributions substantially lags behind the

others, then many MIN-scores will be determined by the bottleneck condition, and the

best correlates of the outcome will be indicators of that barrier.

The final outcome of the use of the RWA model is that it expects both structural

and cultural factors to emerge as correlates. The RWA specification leaves little room

for disciplinary debates of the type “economics versus culture” (cf. Lesthaeghe 1998),

or, by extension, for squabbles between economics, sociology, or political science.



Any one of these three can come up with strong “correlational” results, but the irony

is that “victory” for a discipline can be claimed following the identification of a type

of regressor (e.g., economic, structural, cultural…) with the largest and most

significant coefficients, when in fact such statistical predictors could merely identify

the slowest moving condition in an innovation and diffusion process4. The earlier

types of analysis with “the buck stops at socio-economic structure” in sociology and

social history and the subsequent “cultural turn” in the social sciences (see Sewell

2005, esp. chapters 1 and 2) just lead from one form of reductionist fallacy to another.

The RWA model simply recognizes that processes of social change are the outcomes

of (i) socio-economic structures with their specific configurations of opportunities and

constraints AND (ii) of the adaptive capacity of cultural scripts of legitimation, AND

(iii) of policies affecting the technical and legal environments. The AND is again the

logical conjunction, and the factors that cause leads and lags over time can vary

widely. Some configurations are remarkably recurrent ones, but others can indeed be

totally “historically idiosyncratic”.

3. From the FDT to the SDT: The Belgian example revisited

There are several reasons for revisiting the issue of historical spatial continuity in

Belgium. The first one is that we now possess many more SDT indicators than could be

used in the 2002 Lesthaeghe and Neels article. We owe this to Neels’ fertility

reconstructions of 2006 for both cohorts and cross-sections on the basis of the 1990

census, and to the Gadeyne et al. reconstruction of cohabitation trends and differentials

starting from a retrospective question in the 2001 census. The second reason is that

Neels (2006) showed that the Belgian SDT has two components with different spatial

patterns: marriage postponement and premarital cohabitation followed the older

historical pattern of demographic innovation, but fertility postponement did not. Also,

in the US there was a partial disconnection in the spatial incidence of the fertility

postponement and the non-conformity dimensions of the SDT, respectively (Lesthaeghe

and Neidert 2006). More specifically, the states in the lower half of the two

distributions were the same, but those at the vanguard differed: the North Atlantic states

were the leading ones with respect to fertility postponement, whereas several western

and Pacific states were leaders with respect to cohabitation.

4 An example will elucidate this point. Suppose that everyone satisfies the R condition, but many would not

satisfy the W and/or A conditions. The correlates of the outcome would be connected to W and A and not to

R, since this is a constant. The fact that the R condition is already satisfied is of paramount importance for the subsequent development of the process (necessary condition), but indicators of R will not be significant in

any correlation analysis, and they are likely to be dismissed as “unimportant”.



Before turning to statistical analyses of the Belgian data, a short digression is

necessary to elucidate the secularization history of the country. This history contains a

few features that are essential to the understanding of the role of the W-factor and its

relation to socio-economic structural determinants.

3.1 The three secularization waves in Belgium: 1750-1970

As in much of Western Europe, the first signs of secular, rationalist thinking emerged

with the Enlightenment during the second half of the 18th

century. It was essentially an

elite phenomenon (e.g., in Freemason lodges), but capable of being one of the driving

forces of the Brabant Revolution of 1789, or the first attempt at establishing Belgian

independence. This first wave of secularization gained much wider popular support

during the French Revolution, and many urban parishes, especially in Liège, failed to

observe the marriage ban during the so-called “closed periods” of Lent and Advent.

Also, during the Napoleonic period and the period of reunification with the Netherlands

(1815-1830), Catholicism remained under state control, but this did not imply that the

secular ideas had penetrated in all regions of the country. As became very clear at the

time of the Belgian independence in 1830 and with the concomitant Catholic

restoration, numerous areas in the Dutch-speaking northern half of the country had

remained staunchly loyal to Catholic clergy and doctrine.

The early map of this first secularization wave can be constructed on the basis of

the degree of non-observance of the marriage ban during Lent and Advent, i.e. the

period of 40 days before Easter and the similarly “closed” period prior to Christmas

(see maps 1, 2, and 3 for quintiles). The Napoleonic Civil Code had made a civil

marriage obligatory and a church marriage optional. This remained so after Belgian

independence, and hence the Catholic marriage ban was no longer an impediment for

freethinkers or for couples in need of a “shotgun” wedding. The index used here is

simply the percentage of marriages in March (approximation for Lent) and December

(idem for Advent) divided by 2/12, or the proportion that would be observed without

marriage seasonality. Hence, this MLA index equals 100 if no marriage ban is being

observed, and becomes a much smaller number if the Church rules are being respected.

For instance, during the last decade of the Austrian occupation the MLA index often

comprised between 5 and 15 in Flemish parishes and around 20 in Walloon ones,

meaning that the number of marriages was less than one fifth of the expected number

for two months in the absence of seasonality. After the French takeover these indices

typically increase to 30 - 50, and even exceed 100 for a few years in Liège (Lesthaeghe

1991: 276-279).



Map 1: Marriages during Lent and Advent (MLA-index) in Belgian

arrondissements, 1841-1847 (quintiles)

As the MLA map with quintiles shows for the period 1841-1846, the impact of the

first secularization wave is markedly stronger for the Walloon or francophone

arrondissements (see appendix, map A 1, for their identification) of the southern half of

the country than for the Flemish ones to the north. Even more strikingly, the degree of

non-observance of the ecclesiastic marriage ban is pronounced in many strictly rural

Walloon arrondissements, with those of Arlon, Bastogne, Virton, and Neuchâteau

having higher MLA values than the emerging industrial arrondissements of Charleroi

and Mons. Similarly, other Walloon rural arrondissements match the Charleroi value,

and these are Waremme, Nivelles, and Philippeville. A correlate of this rural

secularization in Wallonia is the strong allegiance to the Liberal party, which did not

only represent the freethinkers among the higher bourgeoisie, but also those among the

wealthier artisans and farmers. By contrast there was no such strong rural Liberal

support in Flanders, and only in the larger towns of Antwerp and Ghent was there a

militant secular presence. A major point of dispute in the initial Belgian “two pillar”



system (Catholic versus Liberal) was the staunch competition between the Catholic

schools and the secular state or municipal schools. Jesuit and episcopal colleges of

secondary education and the Catholic University of Louvain (reopened in 1830)

produced the elites for the Catholic pillar, whereas the State universities of Ghent and

Liège and the small Free(thinker) university of Brussels remained secular strongholds.

After 1860 a third “pillar” was added, corresponding to the rising importance of

the Socialist Party. This generated a second secularization wave in all major industrial

and urban areas as well. The MLA map for 1860-1865 clearly shows how the three

major Walloon industrial arrondissements (Mons, Charleroi, Liège) and the

arrondissement of Brussels have MLA values in excess of 60, which is already

indicative of a major weakening of the ecclesiastic ban. By 1881-1884 only five

Walloon arrondissements have values below 60, whereas there are only two Flemish

ones with values above 60.

Map 2: Marriages during Lent and Advent (MLA index) in Belgian




The double origin of secularization in Belgium before 1900 not only implies a

leading position for Wallonia, but also that both rural and industrial or urban

arrondissements are present in the upper half of the secularization distribution. In other

words, there is only a modest positive correlation between secularization and

industrialization/urbanization, which is a statistical bonus (low multicollinearity) when

it comes to measuring the separate effects of secularization versus

industrialization/urbanization upon, for instance, the speed of the marital fertility

decline (Lesthaeghe 1977: 196-220; Lesthaeghe and Wilson 1986: 261-292).

Map 3: Marriages during Lent and Advent (MLA index) in Belgian


Until the 1960s the map of secularization remains very stable in Belgium, as can

be gleaned from the correspondence between the MLA map for 1881-1884, the map of

the secular vote (% Socialist + Communist + Liberal) in 1919 (first elections based on

universal male suffrage), the map for the secular vote in 1958, i.e. at the time of the last

1881-

84



“school war”5, and the 1964 map of percentages of adults absent during the annual

Sunday Mass census (see maps 4, 5, and 6). Noteworthy in this series of secularization

maps is that the least secularized arrondissements are all located in Flanders and form

two clusters: a western one in the province of West Flanders (arrondissements of Ypres,

Diksmuide, Tielt, Roeselare) and an eastern one comprising the whole of the province

of Limburg (arrondissements of Hasselt, Maaseik, Tongeren) and the adjacent Campine

area (arrondissement of Turnhout). Moreover, as can be seen in Map 2, these two

persistently Catholic regions were already in evidence from the 1860s onward. Equally

noteworthy, however, is that the eastern cluster of Flemish arrondissements remained

strongly attached to the Catholic Church till the 1960s, despite its industrialization after

World War I, i.e. along the Antwerp-Liège axis (Albert canal) and the Limburg coal

fields.

Map 4: Secular vote for liberal, socialist, and communist parties,

1919 (quintiles)

5 The 1958 “school war” erupted as a result of the Liberal-Socialist government’s attempt at restructuring the subsidy conditions of the Catholic school system. At that time many villages only had Catholic primary

schools, and the government then also wanted to create secular counterparts.



Map 5: Secular vote for liberal, socialist, and communist parties,

1958 (quintiles)

The third secularization wave starts during the 1960s and corresponds to the

disappearance of the hitherto Catholic strongholds in the northern half of the country.

From then onwards Flanders catches up with Wallonia, and the marked contrast

between these two regions, formerly coinciding with the linguistic border, becomes less

pronounced. This is also the period of political “depillarization”, with the growth of

alternative parties such as the Greens and later on the regionalist and populist right also

(e.g. Vlaams Blok, Front des Francophones). During the late 1960s and 1970s all

sources of authority are being questioned, and this applies to political parties, the

church, the university system, the army, and the judicial system alike. Not only does the

period 1965-1975 correspond to a major breakthrough of the “post-materialist” and

expressive values supportive of female emancipation and the sexual revolution, but this

is equally the era of massive female educational progression to full secondary education

and beyond.



Map 6: Sunday mass absenteeism among the adult population (15+),

1964 (quintiles)

After 1975 the spatial secularization differentials are largely played out, but two

centuries of spatial contrast in this respect still lingered on with respect to everything

with a moral or ethical dimension. In other words, a set of regional subcultures along a

conservative - liberal dimension still continued to be operative, despite the fact that the

more strictly religious dimension is no longer a major part of it. On the other hand, the

rise of female higher education and of female labor force participation outside the

domestic sphere created a new dimension, equally springing up from the 1960s onward.

As we shall illustrate in the next section, these two aspects will also be of major

relevance for the unfolding of the various demographic characteristics of the SDT.



3.2 Spatial continuity: Indicators and covariates for the Belgian example

To illustrate the spatial continuity over a period of almost a century and a half, we will

resort to a series of both FDT and SDT indicators for the 41 (and later 43) Belgian

arrondissements. The correlation matrix for these demographic indicators is

subsequently analyzed via a classic Principal Component Analysis6, which extracts

three orthogonal factors (Varimax rotation) that jointly account for 80% of the total

variance contributed by the entire pool of indicators. The next step consists of linking

these demographic factors to a series of covariates of both a structural and a cultural

nature in order to identify the best correlates.

The subset of demographic indicators related to the FDT contains all the Princeton

indirectly standardized measures of marital fertility (Ig), of proportions married (Im),

and of non-marital fertility (Ih) (Coale 1965: 207), computed for all the census dates

between 1880 and 19707. In addition, we also use the percentage of the total marital

fertility decline that had already been completed by 1910 (delta Ig 1880-1910,

Lesthaeghe 1977:109). Equally belonging to the FDT, but to its later years, are

measures for the cohort of women born in 1931-1935, since the events of interest would

typically have taken place in the late 1950s and the 1960s. For this cohort use is made

of their mean age at first birth, their percentage of non-marital births, and their

6 In all instances where Principal Component Analysis (PCA) is used, axis rotations are orthogonal, and the

resulting factors are uncorrelated dimensions. All factors (= principal components in our case) with

eigenvalues larger or equal to unity were retained. We use the terms “principal component”, “factor” and “dimension” interchangeably. The term “factor” loading refers to the correlation coefficient between the

factor and the indicator variable. Factors are composite measures or underlying dimensions of a set of

indicators. Factors are constructed in such a way that they have means of zero and standard deviations of unity. Maps that present these factors use standard deviations as units of measurement, and divide the scale in

quintiles. 7 The Princeton indicators are all indirectly standardized measures, because of the absence of numbers of

births broken down by age of the mother in most historical series of European countries. The Princeton

indices all use Hutterite marital fertility schedules by age, h(a), to capture “natural fertility”, i.e. very high

fertility in the absence of contraception. The four indices are: * Index of marital fertility Ig = observed legitimate births / Sum M(a).h(a) where M(a) is the number of

married women by age, typically recorded in censuses. Age groups start at 15 and end at 49.

* Index of overall fertility If = observed births (all)/ Sum W(a).h(a), where W(a) are all women by age. * Index of illegitimate fertility Ih = observed extra-marital births/ Sum U(a).h(a), where U(a) are unmarried

women by age.

* Index of proportions married Im = Sum M(a).h(a)/Sum W(a).h(a), or proportions married weighted by Hutterite fertility.

With these indices If = Ig.Im + Ih(1-Im) or in the instance of very low extra-marital fertility, If = Ig.Im. This

allows for a simple decomposition of overall fertility If into a marital fertility and a nuptiality component

respectively.

The “speed” of the fertility decline (Lesthaeghe 1977) is measured as the decline of Ig between the value Ig

(t) at the onset and a value at a later date Ig(t+x), divided by Ig(t) -.300, where .300 is taken as the “end” value of the transition. The measure gives an idea of the proportion of the total marital fertility transition

already covered between the onset of the process and a later date.



percentages of ever divorced. Finally, four more indices, all measured in 1960-1962,

pick up relevant information for the end of the FDT: the total first marriage rate

(TFMarR), the total fertility rate (TFR), the mean age at first birth, and the percentage

of the TFR that was realized after age 30 (capturing mainly higher order parities and

unplanned births during the pre-pill era).

The subset of SDT demographic indicators pertains to the rise of divorce, the

postponement of marriage and fertility, and the rise of cohabitation and of non-marital

births (mainly among cohabitants). For the periods 1969-1971, 1979-1981, and 1998-

2000 use is made of the classic total fertility rates (TFR), the mean age at first birth, and

the percentage of the TFR occurring after age 30. For the years 1970, 1981, 1991, and

1999 we possess the total first marriage rates TFMarR and the total first cohabitation

rates TFCohabR. The latter could be computed from the 2000 census information on the

year that first premarital cohabitation was initiated and the year of birth of the female

respondents (see Gadeyne et al. forthcoming). Further information on cohabitation is

available in the form of percentages cohabiting among women aged 20-24 and 25-29

measured in the 1990 census, and also as an indirectly age standardized index for 2000,

along with similar indices for proportions married and divorced (Deboosere et al. 2009).

Furthermore, information has been added pertaining to the cohort of women born in

1961-1965 and mainly occurring in the 1980s and 1990s: age at first birth, percent non-

marital births, and proportions ever divorced (Neels 2006)

The socio-economic covariates essentially capture the processes of

industrialization and urbanization, the degrees of literacy, and, for the second half of the

20th

century, the rise of female secondary and higher education, along with their labor

force participation. More specifically, for the periods 1880, 1890, 1900, and 1910 use is

made of the percentages of the male labor force in agriculture (and for 1910 also in

agriculture plus cottage industries), an index of urbanization and industrialization, and

the literacy rate for the population aged 15-55 (Lesthaeghe 1977 160 ff.). This series

continues with the urban percentages from 1920 to 1970. For the female cohorts born in

1931-1935 and in 1961-1961 we have the proportions ever worked. For the older

cohort, educational achievement is measured as the percentage at least completing full

secondary education, and for the younger as the percentage having a post-secondary

degree (Neels 2006: 87-88, 188-189).

On the “cultural” side we have a series of measures of secularization and of

linguistic homogeneity. The oldest series of secularization measures are the indices of

marriages during Lent and Advent (MLA) for 1841-1847, 1860-1865, and 1881-1884

(Lesthaeghe 1991: 271). The next pair is the percentage votes for secular parties

(Liberal, Socialist, Communist) in 1919 and 1958, and the last measure is the

percentage absent from Sunday Mass in 1964. Linguistic homogeneity is measured as

the percentage of the population over age 15 that only speaks the language or dialect of



the region (monolingual - Dutch or Flemish in the North, French or Walloon in the

South). A low degree of language homogeneity captures the presence of linguistic

minorities in a given region. These were either the original population (e.g., Flemings in

Brussels, German speakers in the arrondissements of Bastogne and Verviers) or

immigrants (e.g., Flemings in Wallonia). The use of languages was no longer recorded

after World War II, as being too sensitive politically.

3.3 Statistical results for the Belgian example

The statistical analysis consists of two steps. First, the set of demographic indicators

mentioned above is being reduced to a much smaller set of dimensions (PCA, Varimax,

orthogonal factor rotation). Second, the best social and cultural correlates of each of the

demographic dimensions are identified. This gives a succinct description of the

underlying structure of the entire correlation matrix.

The three dimensions are clearly identifiable and tell the three basic stories.

Factor 1 is the long-term continuity dimension of demographic innovations. This

dimension identifies the leading and lagging regions with respect to fertility control and

contraception during the FDT. But it continues to reflect a subset of SDT indicators,

and more particularly all those associated with the weakening of the marriage

institution: the rise of divorce, postponement of marriage, increasing cohabitation, and

parenthood within the cohabitation context.

Factor 2 reflects the historical rise and later decline of “traditional” non-marital

fertility. This factor no longer has demographic indicators after the 1960s, which means

that non-marital fertility from then onwards takes place in an entirely different context.

This discontinuity essentially reflects the shift away from illegitimacy of non-marital

births, not legalized by shotgun marriages and occurring to single women or adulterous

married women, to extra-marital births mainly among cohabitants.

Factor 3 has no clear deeper historical roots, but is a novelty typical of the SDT:

postponement of parenthood among all types of couples. As already shown by Neels

(2006), the SDT in Belgium has two separate components, one related to divorce,

marriage, and cohabitation, and one related to postponement of parenthood. These two

components have different geographies. The former reflects the long-term innovation

dimension (see Factor 1), but the latter has a geography of its own.

We shall now proceed with a more detailed investigation of each of these three

dimensions.



3.3.1 Factor 1: Successive demographic innovations and long term spatial

continuity

Table 1 presents the long-term continuity dimension of innovation. High scores on this

dimension for arrondissements reflect the presence of a leading position throughout the

whole period from the 1870s until 2000, whereas low scores identify the

arrondissements that were slow in adopting the innovations of both FDT and SDT. The

left side of the table contains the best indicators of dimension one with their factor

loadings (or correlation coefficients) of 0.700 or better. The right hand panel lists all the

structural and cultural correlates of dimension 1 with correlation coefficients of 0.600

or better.

The prime continuity dimension exhibited in Table 1 has factor loadings on an

impressively long series of demographic indicators related to the FDT: all indicators

related to the historical marital fertility control, the departure from the restrictive

Malthusian marriage system, and even the continued higher fertility past age 30 in the

pre-pill early 1960s. The series then continues with indicators pertaining to the SDT;

the rise of divorce, but above all the postponement of marriage in favor of premarital

cohabitation. And by the 1990s procreation among cohabitants has joined the set as

well. Note, however, the absence of the measures of fertility postponement.

Aside from the 19th

century indicators of proportions of the male labor force in

agriculture and cottage industries, all the best correlates of the demographic spatial

continuity dimension are indicators of progressing secularization, already starting with

the MLA index of 1841-1846 and continuing uninterrupted until the 1960s with the

percentages absent from Sunday Mass. Also note that indicators of urbanity or

industrialization after 1920 are not in this set. Within the RWA framework this is

strongly indicative of the fact that it was the W-condition that was the limiting one. In

other words, essentially the moral and religious objections to two entirely new forms of

behavior, i.e. controlling fertility through contraception and replacing marriage by

cohabitation, were constituting the decisive limiting conditions that shaped the

geography of these successive demographic innovations.

The map of the factor scores (expressed in standard deviations) of the

arrondissements on the demographic continuity dimension is presented below (map 7).

High scores are indicative of a leading position. Map 7 clearly shows that Walloon

arrondissements, along with Brussels, were systematically at the vanguard with respect

to all innovations captured by Factor 1. The language border is a well-demarcated

barrier (see also Lesthaeghe 1977: 111-114). Moreover, many Walloon rural

arrondissements are in the first quintile (cf. Philippeville, Dinant, Marche, Huy,

Nivelles, Tournai, Ath), whereas the major Walloon industrial poles (Charleroi, Mons,

Liège) are not.



Table 1: Long term continuity with respect to demographic innovations in

Belgian arrondissements, 1841 – 2000 (principal component analysis

results -- Factor 1)

Demographic Indicators with Factor

Loadings GE .700

Best Social Correlates of Factor 1 with Corr.

Coeff. GE .600

1880 Index of Marital Fertility Ig -0.881 1841-46 Marriages in Lent & Advent

MLA 0.645

1890 Ig -0.898 1860-65 MLA 0.780

1900 Ig -0.944 1881-84 MLA 0.821

1910 Ig -0.938 1890 % in Agriculture -0.768

1880-1910 Speed of marital fertility

decline 0.849 1900 % in Agriculture -0.625

1920 Ig -0.879 1910 % in Agric. + Cottage Ind. -0.652

1920 Index of prop. Married Im 0.709 1919 % Secular Vote (Soc +

Com + Lib) 0.897

1930 Ig -0.784 1958 % Secular Vote (Soc +

Com + Lib) 0.772

1947 Ig -0.710 1964 % Non-attendance Church 0.698

1960-62 % TFR realized after age 30 -0.842 1980s % Women ever-worked in

cohorts 1961-65 0.672

1960-62 Total 1st Marriage Rate -0.706

1961 Index of Non-marital fert. Ih 0.716

1961 Total 1st Cohabitation Rate 0.850

1960s % Women ever Divorced in

cohorts 1931-35 0.705

1960s % Non-marital 1st births in

cohorts 1931-35 0.732

1967-70 Total Divorce Rate 0.764





1990 % women 20-24 in

Cohabitation 0.764

1990 % women 25-29 in

Cohabitation 0.840


2000 Index Cohabitation 0.803

2000 Index 1st Marriages -0.902

2000 Index Divorce 0.750



Finally, the map also clearly shows the two Flemish regions that were

systematically at the tail end of the innovation distribution. These correspond entirely

with the zones that had the longest resistance to the first two secularization waves, i.e.

the West Flemish zone with Ieper, Diksmuide, Tielt, and Roeselare, and the Limburg-

Campine zone with Hasselt, Maaseik, Tongeren, and Turnhout (see maps 1 through 6).

Map 7: Factor 1 (quintiles): Long-term continuity dimension in FDT and

SDT innovation in Belgian arrondissements, 1880-2000

3.3.2 Factor 2: Historical illegitimacy in fertility and the urban-industrial

connection

The second dimension emerging from the Principal Component Analysis mainly

captures the remarkably stable spatial pattern of out-of-wedlock fertility in the period

1880-1930 (see Princeton Ih index), and later on the presence of higher proportions



married (Im) but lower fertility (see Table 2). Also noteworthy is the fact that there are

no demographic indicators measured after 1980 loading on this factor. Hence it is fair to

state that this dimension is more a historical one and unrelated to the SDT. Equally

remarkable is that, also on the side of its correlates, only indicators of urbanity are

identified up until 1970. Finally, note that the negative correlation with adult literacy in

1880 stems from the fact that the Belgian industrial poles of the 19th

century, but to

some degree also the larger urban ones, were attracting illiterate immigrant populations.

But after World War II this negative relationship between literacy and urbanity was

being reversed in tandem with the growth of the tertiary sector in the economy.

Table 2: The historical dimension of non-marital fertility in Belgian

arrondissements, 1841-2000 (principal component analysis results -

Factor 2)


Loadings GE .700


Coeff. GE .600

1880 Index non-marital fertility Ih 0.837 1880 % literate -0.621

1890 Ih 0.891 1890 % urban 0.711

1900 Ih 0.923 1900 % urban 0.756

1910 Ih 0.901 1910 % urban 0.790

1920 Ih 0.931 1910 % literate -0.633

1930 Ih 0.838 1920 % urban 0.720

1947 Index of non-marriage Im 0.713 1930 % urban 0.726

1970 Im 0.785 1947 % urban 0.694

1970 Index of marital fertility Ig -0.796 1950s cohort 1931-35 with higher

sec educ 0.628

1969-71 TFR -0.773 1970 % urban 0.685

1979-81 TFR -0.734

Map 8 shows the geography of this second dimension. It strongly reflects the

degree of urbanization and industrialization with high values for the Brussels – Antwerp

axis in Flanders, and for the Hainaut industrial belt in Wallonia (Mons, Charleroi). But

two other less industrialized arrondissements are also in the top quintile: Ostend on the

Flemish coast and Thuin adjacent to the Hainaut industrial belt. By contrast, all the

highly rural Walloon arrondissements of the Ardennes and several in the West Flemish

rural belt are typically gathered in the lowest quintile. These were also arrondissements



with historically negative migration rates, a feature that contributed to the strengthening

of their rural character. Finally, it is impossible not to notice that the language border

plays no role here whatsoever.

Map 8: Factor 2 (quintiles): Pre - World War II “illegitimate” fertility in

Belgian arrondissements and its historical urban-industrial

connection

3.3.3 Factor 3: SDT fertility postponement

The third factor identified in the Principal Component Analysis pertains exclusively to a

feature that belongs to the SDT: fertility postponement and differential catching up of

fertility after age 30 (see Table 3). This is being indicated by the positive correlations

with the mean ages at first birth from 1970 onward, and the percentage of the total

fertility rate (TFR) realized after age 30 from roughly 1980 onward. Areas with high

scores on this third dimension are then typically those with stronger fertility



postponement and subsequent fertility recuperation at later ages. Not surprisingly, these

areas are also identified by two structural features: high post-secondary female

education and high female employment rates for the cohorts born in the 1960s.

Table 3: SDT fertility postponement dimension in Belgian arrondissements

1950-2000 (principal components analysis results - Factor 3)


Loadings GE .700


Coeff. GE .600

1950s % of 1st births prior to age

25, cohort of 1931-35 -0.773 1980s

% women with higher

education, cohort 1961-

1965

0.807

1969-71 Mean age at 1st birth 0.701 1980s % women ever in labor

force, cohort 1961-1965 0.601

1979-81 Mean age at 1st birth 0.928

1979-81 % of TFR after age 30 0.844

1989-91 % of TFR after age 30 0.740


1998-00 % of TFR after age 30 0.809


Map 9 below shows the geography of the fertility postponement dimension. Along

with the capital Brussels, all major Flemish urban areas and economic growth poles are

represented in the top quintile (Halle-Vilvoorde, Leuven, Antwerp, Ghent, Bruges), in

tandem with the better performers in Wallonia (Nivelles adjacent to Brussels, Verviers,

and Arlon with its large population employed across the border in the Grand Duchy of

Luxemburg). Much earlier fertility is still maintained in the whole of southern Wallonia

along the French border, but particularly in the Hainaut old industrial belt. The same

also continues across the language border in the southern part of West Flanders.

However, the Flemish area of less fertility postponement has fewer cohabitants and

much lower fertility rates among such cohabiting couples than its Walloon counterpart,

as the stark differentiation in Factor 1 had already indicated.



3.4 The Belgian experience in a nutshell

The FDT in Belgium has two main components: (i) control over the level of marital

fertility and (ii) the historical bulge and subsequent decline in “illegitimate” fertility.

Map 9: Factor 3 (quintiles) -- the fertility postponement aspect of the SDT in

Belgian arrondissements

The best spatial correlate of the former was the degree of secularization, pointing at the

fact that the bottleneck condition in the RWA trio must have been the W-condition or

the religious and moral acceptability of contraception. The Belgian experience is

furthermore one of considerable continuity, since the secularization dimension

maintains a stable spatial pattern until at least the end of the 20th

century, and continues

to condition the spread of the “non-conformist” aspects of the SDT (i.e. cohabitation,

parenthood among cohabiters) as well. This implies again that the moral acceptability

or the W-condition continued to be the limiting condition for this “innovative” aspect of

the SDT as well. This is not surprising, given the fact that both contraception (FDT) and



cohabitation (SDT) were in the respective time periods viewed as “unconventional” or

even “immoral”.

The contrast is provided by the other aspect of the SDT; i.e. the fertility

postponement component. This feature exhibits a totally different geographical pattern,

with female advanced education and concomitant employment emerging as the best

spatial correlates. Hence there is no trace here of the operation of the W-condition, and

this is quite logical since postponing parenthood is not behavior that directly violates

moral or religious prescripts. As a consequence, the geography of fertility

postponement mirrors the R-condition, since better educated and employed women

were typically in the vanguard of the postponement transition (cf. Neels 2006, for

individual level data).

4. From FDT to SDT: The Spanish experience

The comparison of Belgium and Spain allows for a more revealing study of cultural and

linguistic factors involved in the FDT and SDT. Both countries share a Catholic

tradition and have similar roots of secularization (Enlightenment, Marxism), and both

are linguistically heterogeneous. Yet in a host of other characteristics they are quite

different. To start with, Belgium was the first continental country to move through the

Industrial Revolution, whereas Spain was near the tail end of the distribution in this

respect. In fact only two Spanish areas had an earlier 19th

century industrial

development, Catalonia with textiles and the Basque country with metal industries.

Secondly, Spanish landholding systems were characterized by the juxtaposition of

“minifundios” of less than one hectare and permitting subsistence farming only, and

“latifundios” or large agricultural enterprises operated by day laborers. Despite

agricultural reforms, this duality survived well into the 20th

century, and it is one of the

causes of another major distinction, i.e. political polarization. Indeed, despite a

grounding of politics in similar religious and ideological divisions, political history is

far more turbulent in Spain than in Belgium. Starting with the guerilla war against

Napoleon there was a steady growth of polarization between Catholic and secular

factions, culminating in the Spanish Civil War. Thereafter both cultural and technical

developments were slow for most of the Franco era (1939-1975), i.e. until at least 1965,

whereas the rest of Western Europe was building new industrial and infrastructural

capacities on the ruins of World War II. In fact it is not until the end of the Franco era,

the constitution of 1978, and EU membership that Spain is propelled at high speed into

late 20th

century economic, technical, and cultural modernity.

A number of these points need further elaboration, since they will be of major

relevance to understanding the spatial patterns of the FDT and SDT in Spain.



4.1 The Spanish spatial secularization pattern

Just as in Belgium (then known as the Austrian Low Countries), the start of

secularization in Spain is linked to the Enlightenment and to the “enlightened

despotism” of the last two 18th

century monarchs (Carlos III and IV). During their reign

the Jesuits were expelled (1767), the church was forced to sell off land and loan the

proceeds to the state, and the private organizations to stimulate economic,

philosophical, and scientific development were officially sponsored. These

organizations, called Sociedades Economicas de Amigos del Pais, received legal

licenses to foster their activities both in Spain and the colonies. After the guerilla war

with the Napoleonic forces, Spain adopted a liberal constitution in 1812.

During the first half of the 19th

century Spanish liberalism replaced the Ancien

Régime through political struggles, revolutions, and civil wars. After the first Carlist

war in 1840 and the traditionalists’ defeat, the church’s role in politics, economics, and

social works was further reduced. By 1843 many episcopal sees were vacant, cathedral

chapters were controlled by laymen, most male religious orders were banned, and more

church property had been sold to service the public debt. Moreover, the tithes were

abolished, priest became civil servants, and there was no curtailment of anticlerical

urban riots.

Under the reign of Isabel II (1843-1868) and later under that of her son Alfonso

XII (1874-1885) and during the dictatorship of Primo de Rivera (1921-1930) the church

gradually recovered. During that period the main interruption was the First Republic of

1873-1874, but that only constituted a brief intermezzo. However, by the end of the 19th

century there is a further polarization of the political spectrum. Not only were the

Socialists present on the Left, but so also were more radical factions with anarchist

philosophies. Similarly, fascists enlarged the extreme Right, hitherto mainly made up of

the strongly Catholic Carlists. Hence, at the turn of the century the extremist forces

were growing that led to the events of the 1930s, i.e. the Second Republic and the Civil

War.

A first glimpse of the regional pattern of these divisions can be obtained from the

index of marriages in Lent (ML-index) for 1900 and 1901. In contrast to Belgium,

marriages during Advent were frequent in Spain, and only those in Lent were banned or

discouraged. The ML index for Spain is then simply the percentage of marriages in

March divided by 1/12 (assuming a flat distribution of marriages within the year), or

simply multiplied by 12. The provincial values at that time span the complete spectrum,

from virtually no such marriages (ML values below 25) to almost a complete non-

adherence to the practice of not marrying during the ecclesiastic “closed period” (values

above 80). The strongly Catholic areas of Asturias and the provinces of Castile and

Leon clearly show up as white zones on the map of the ML-values for 1900-1901 (Map

10). Maps with other indicators, such as the number of inhabitants per clergyman in the



census of 1887 (Reher et al. 1993), or the number of inhabitants per seminarist in

dioceses in 1963, show almost exactly the same areas as being Catholic strongholds. At

the other end of the distribution there is a large north eastern zone of strong

secularization, made up of Catalonia and adjacent Huesca and Zaragoza plus Rioja, and

a large southern zone covering Andalusia (with the exception of Cordoba) and adjacent

Badajoz, Ciudad Real, and Alicante. Also the western half of Galicia (A Coruña,

Pontevedra) distinguishes itself by higher levels of non-adherence to the marriage ban

than its Catholic hinterland. Finally, the Canary Islands equally fall in the top quintile

of non-observance of the marriage ban.

Map 10: Index of marriages during Lent (ML) in Spanish provinces

(quintiles), 1900-1901

Further into the 20th

century other indicators of secularization were constructed.

For the Second Republic use was made of the percentage in each province of the

elected diputados for the Left in 1931 and 1936 and for the Right in 1933 and 1936.

Centrist diputados were left out in setting up these indicators. For the post-Franco era

the indices are the percentage vote for the Left in the 1977 election (including regional

leftist parties) and the percentage of marriages that were only civil marriages in 2001

and 2009. Together with the ML-index for 1900-1901, these seven indicators were

Index of Marriages in Lent

7.0 - 26.0

26.0 - 46.0

46.0 - 66.0

66.0 - 80.0

80.0 - 96.0



subjected to a Principal Component Analysis to extract the underlying orthogonal (i.e.

uncorrelated) dimensions. It turns out that there is a clear distinction between a

historical spatial dimension of “old secularization” and a newly emerging post-Franco

“new secularization” with a different geography. The best indicators of the older pattern

are, not surprisingly, the percentage of diputados for the Left in 1936 (factor loading or

r = .853), for the Right in 1936 (-.757), percentage vote for the Left in 1977 (.704), the

ML-index in 1900-01 (.671), the percentage of diputados for the Left in 1931 (.559),

and for the Right in 1933 (-.505). The best indicators of the “new secularization”

dimension are the percentages of civil marriages in 2009 (.947), in 2001 (.942), and the

percentage of diputados for the right in 1933 (-.580). It can be noted that the 1977

election results were still correlated with the Second Republic political maps (see also

Linz and Montero 1999), but that the 21st century indicators have become quite

independent of the historical dimension.

The two factors of old and new secularization respectively are presented in Maps

11 and 12. Note that the scale of the principle components (factors) is expressed in

standard deviations, and that the distribution has a mean of zero and a standard

deviation of unity.

The comparison of Maps 10 and 11 show that the historical secularization

dimension in Spain was quite stable until the 1970s. The core of the most Catholic area

is made up of Castile and Leon and the provinces of Castile-La Mancha around Madrid

together with Teruel. This large central zone spreads further out to include eastern

Galicia (Lugo, Ourense), Cantabria (Santander), Navarra, and the northern Basque area

(Bizkaia, Gipuzkoa). The most secular areas are again the whole of the South, the entire

Mediterranean coast (but not the Baleares), inner Catalonia (Lleida), and northern

Aragon (Huesca). Finally, there is a band of northern provinces that is situated in the

central segment of the distribution of the historical secularization factor, and it is made

up of western Galicia (A Coruña, Pontevedra), Asturias, Rioja, Alava, and Zaragoza.

Another striking feature for Spain is the strong connection between Catholicism

and literacy. In 1887 the census figures for male analfabetism show a contiguous zone

of percentages lower than 41% in the provinces of Castile and Leon and along a narrow

Atlantic stretch from eastern Asturias, Santander to western Bizkaia. Intermediate

figures form a circular band around this Catholic heartland to include Galicia and the

northern part of Castile-La Mancha, and also emerge along the Catalan coast. But in

much of the rest of Spain, and notably in the south and along the central and southern

Mediterranean coast, male illiteracy was higher than 70% (Reher et al. 1993).



Map 11: The historical secularization dimension in Spanish provinces

(quintiles of factor values)

The map of 1887 for female illiteracy shows a similar pattern with levels below

two thirds in Castile and Leon, Navara, Rioja, the whole of the Basque region,

Santander, and eastern Asturias. By contrast, levels in excess of 85% of female

illiteracy prevailed in the whole of Galicia and in the entire south-eastern segment of

the country, roughly from Castellon to Almeria and further inland to Ciudad Real. In

contrast to Belgium, where illiteracy disappeared almost entirely after the First World

War, the Spanish low literacy zone persisted much longer (Vilanova-Ribas and

Moreno-Julia 1992). Moreover, the concordance between higher literacy and better

education on the one hand and Catholicism on the other meant that the civil service in

Madrid and the elite of the Franco era were disproportionally overrecruited from Castile

and Leon, thereby fuelling more antagonism between the political rightist center of the

country and the more secular periphery. However, things changed after 1975.

As stated before, Spain underwent a rapid and profound transformation during the

post-Franco era in just about every single domain of life, from economic modernization

and rising standards of living to much more centrist politics and the total disappearance

of the extreme Left and Right alike (Linz and Montero 1999). Obviously Spain had no

Factor score

-1.99 - -1.03

-1.03 - -0.20

-0.20 - 0.55

0.55 - 0.92

0.92 - 1.97



more appetite for either the “nacionalcatolicismo” (Botti 1992) of Franco or for

adventures with the far Left. On the other hand, just like in Belgium, regionalist parties

and movements gained both ground and influence with marked degrees of devolution of

economic power and with linguistic and cultural autonomy as a result.

In tandem with the post-Franco modernization of its economic and social fabric,

Spain also underwent a fast and profound new secularization wave. The proportion of

practising Catholics (Mass at least a few times per month) drops from a plateau of no

less than 90% during the 1950s and 1960s to just over 50% in only the five years

between 1970 and 1975 (Requeña 2005). And from 1975 onward there is a further

linear decline to less than 30% in 2002. The Spanish secularization is not only visible in

the institutional spheres, but equally in profound changes with respect to individual

ethics. For instance, to everybody’s amazement in the rest of Europe, Spain joined the

vanguard countries in recognizing same sex marriages. It comes therefore as no surprise

to witness a steady rise of marriages that have the civil part only and do away with the

religious ceremony. For instance, for Spain as a whole, in 1976 barely 0.3% of

marriages were civil ones only, but by 2000 this had grown to a quarter (24.1%) (INE,

2005), and by 2009 to more than half (54.9%) (INE website). In some provinces this

figure exceeds 60% (Baleares, Barcelona, Girona, Gipuzkoa, Lleida, Las Palmas, Santa

Cruz de Tenerife, Tarragona, and Valencia). Hence, with the disappearance of the old

politico-religious cleavage, a new spatial pattern emerges for the latest secularization

wave. This pattern is caught by the second principal component in our analysis, and it

is shown in Map 12.

The most striking difference between Maps 11 and 12 is that the whole south is no

longer a component of this late 20th

and early 21st century secularization wave. This is a

major factor in Spanish spatial cultural discontinuity. But, on the other hand, the old

Catholic zones of Castile and Leon and Castile-La Mancha remained at the lower end of

the distribution as well, so that there is one large contiguous area of slower than average

“new secularization”, stretching from Burgos to Cadiz (north-south) and from Murcia to

Caceres (east-west).

The other half of the distribution geographically forms an extended “peripheral

arc”, and contains both provinces that have always been at the vanguard of secularism

and a number of newcomers. Obviously Catalonia remains in the former group, and so

does the linguistically related Comunidad Valenciana. The Baleares complete the

linguistically homogeneous part of this “new secularization arc”. The northern band of

provinces along the Atlantic Ocean and the Pyrenees is also well represented in the top

half of the distribution. This includes “familiar” provinces in this respect such as

Huesca, Alava, and western Galicia. A set of relative “newcomers” such as the northern

Basque provinces, Soria, Asturias, and eastern Galicia complete the “arc”. Finally, it

should be noted that Madrid is not among the trendsetters in this latest secularization



wave, but just in the middle of the distribution, and that the Canary Islands continue to

be highly secularized provinces.

Map 12: The new secularization dimension in Spanish provinces (quintiles of

factor values)

To sum up, one of the major differences between Belgium and Spain in recent

secularization history is that the former country displays a high degree of continuity in

the spatial pattern of secularization, whereas the latter exhibits a clear cleavage between

the old secularization pattern and a new one that has emerged since the 1970s. This

distinction is obviously related to the fact that Belgium has been a typical

“consociational democracy” (Lijphart 1968), i.e. governed by political “pillars” which

had to form coalition governments from the very onset in 1830. The country has an

intact track record of negotiated pacts, deals, and compromises between political elites,

and therefore, unlike Spain, less polarization and no open and virulent conflicts along

the secularization dimension.

Factor score

-1.88 - -0.76

-0.76 - -0.39

-0.39 - 0.14

0.14 - 0.89

0.89 - 2.31



4.2 The demographic dimensions of the FDT and the SDT in Spain

As in the Belgian analysis, use has again been made of a large set of demographic

indicators for Spanish provinces, respectively measuring features of the first and second

demographic transitions. The correlation matrix for the total of 24 demographic

indicators was again analyzed by means of a Principal Component Analysis, and this

set of indicators could be reduced to five basic dimensions, three of which capture the

FDT and two the SDT. These 5 dimensions are described in Table 4.

Table 4: Indicators and dimensions of the first (FDT) and of the second

demographic transition (SDT) in Spanish provinces

A. FDT – First demographic transition dimensions

Factor 1. Late and slow

decline of marital fertility,

resistance to fertility control,

1887-1960.

Factor 2: Prevalence of

earlier female marriage, as

opposed to Malthusian

pattern of late marriage and

higher definitive celibacy,

1887-1930.

Factor 3: Historical pattern

of fertility outside wedlock,

1887-1960.

Princeton index of

marital fertility (Ig

1887)

0.78

Princeton index of

proportions married

(Im 1887)

0.94

Princeton index of

illegitimate fertility (Ih

1887)

0.89

Ig 1900 0.84 Im 1900 0.92 Ih 1900 0.92

Ig 1930 0.85 Im 1930 0.85 Ih 1930 0.74

Ig 1950 0.90 Ih 1960 0.58

Ig 1960 0.74

% total marital fertility

transition completed

by 1930

-0.60



Table 4: (Continued)

B. SDT – Second demographic transition dimensions

Factor 4: The new postponement of

marriage and parenthood, 1995-2001.

Factor 5: SDT - Non-conformist family

formation pattern, 2001-2002.

Women not yet in a union at age

25-29, 2001 0.82 Pct unions cohabiting, 2001

0.64

Fertility postponement index (fert

30+/ fert 20-29 ), 2001 0.93 pct births non-marital, 2001 0.78

Women childless 25-29, 2001 0.93 pct children <15 in one-parent

household, 2001 0.45

Women childless 30-34, 2001 0.91 total fertility TFR, 2002 0.65

Total fertility rate TFR, 1995 0.72

The dimensions described above in table 4 clearly show that we are essentially

dealing with five different features. The resistance to fertility control, the pattern of

early marriage, and the historical levels of extra-marital fertility are the three classic

ingredients of the FDT, whereas the postponement of fertility and the rise of “non-

conformist” family formation are the two SDT components.

4.3 Statistical results for the Spanish example

We shall now discuss the maps for each of the five demographic dimensions of the

Spanish FDT and SDT, and make the link with a set of socio-economic and cultural

spatial indicators.

4.3.1 The FDT marital fertility transition

Following up on the earlier studies of J. W. Leasure (1962) and M. Livi-Bacci (1968), a

detailed analysis of the Spanish historical fertility transition has been offered by Reher

and Iriso-Napal (1987, 1989). Much of what follows here on the FDT therefore largely

echos their findings. However, when it comes to a multivariate analysis in section 5,



another strategy will be followed which will essentially eliminate the multicollinearity

problem that affected the outcomes of earlier studies.8

Map 13 depicts the spatial pattern of the Spanish marital fertility transition. The

factor captures the marital fertility levels as measured through the Princeton Ig measure,

from 1887 through 1960, and the relative speed of the decline of marital fertility

between 1887-1900 and 1930. For the last indicator, the Ig-levels at the onset were

chosen to be either those for 1887 or for 1900, depending on which was highest. This

choice was made in order to accommodate the fact that several provinces still had a

fertility increase between 1887 and 1900. Furthermore, we take 1930 as a good

approximation of the mid-transition point, since the 1920 Ig levels, as used by Reher

and Iriso-Napal, are still very high. For instance, for the rural parts of provinces the

mean Ig level was .66 in 1887, .68 in 1900, and still .62 in 1920. (Reher and Iriso-Napal

1989: 408). For Spain as a whole these values were respectively .65, .65, and .59. The

value for 1930 is .54, which is still fairly high but about the halfway value between Ig =

.65 in 1900 and .40 in 1960. (Coale and Treadway 1986: appendix table: 145).

Aside from Madrid, one major set of contiguous provinces stands out as the

Spanish region with an early fertility decline. This zone comprises the entire Catalan-

Valencian linguistic area, i.e. the three Catalan provinces, the three provinces of the

Comunidad Valenciana, and the Baleares Islands. The three adjacent provinces of

Aragon are also part of the faster fertility transition zone. In the top quintile we also

find the two Canary Islands provinces. Spearheading the marital fertility decline were

the provinces of Barcelona, Tarragona, and Castellon. Taking an Ig value of .30 (or 30

percent of Hutterite fertility) as the endpoint of the transition, these three provinces had

already covered two thirds of the entire marital fertility transition by 1930.

The areas belonging to the latest and slowest quintiles are all located along a north-

south axis stretching from Santander (Cantabria) to Badajoz, and comprising all of

Castile and Leon and Extremadura. Navarra is also part of this slower zone, and so are

Ciudad Real, Jaen, and Cordoba in the south. The following provinces had not even

completed 10% of the transition by 1930: Avila, Burgos, Cordoba, Cuenca, Granada,

Navarra, Salamanca, Segovia, and Zamora.

The best correlates of a fast marital fertility transition (the opposite to what is

shown in the map above) were the following: the percentage employed in industries in

1900 (r = .425), the urban percentage in 1900 (.309), the percentage of diputados for the

left in 1936 (.316) and for the right at the same elections (-.359). As expected, these

correlations are all very modest, and that perfectly echoes the findings of Leasure

(1962) and of Reher and Iriso-Napal (1989). In the analysis of the latter authors other

8 The articles by Reher and Iriso-Napal present both zero-order correlations (r) and standardized regression coefficients (beta). This is a very fortunate decision since the comparison of values of r and beta is very

instructive in assessing the seriousness of the multicollinearity problem.



variables have also been added, such as infant and childhood mortality rates, the

presence of day laborers, and a measure of net migration. In accounting for the Ig-

values of rural parts of provinces in 1920, they find correlations for these three extra

indicators of, respectively, .304, -.341 and -.511.

Map 13: FDT - late and slow decline of marital fertility, 1887-1960

The positive correlation between higher marital fertility and higher infant and

childhood mortality can be explained, as done by the authors, by the impact of

breastfeeding differentials. Shorter durations of breastfeeding lead to both less spacing

of births and higher fertility and to higher mortality prior to age 5. A breastfeeding map

for Spain could further elucidate this point.9 The negative correlation with day laborers

essentially captures the lower marital fertility in a number of southern provinces with

agricultural structures dominated by latifundio. The negative correlation between net

migration and marital fertility in rural areas points in the direction of selectivity, with

traditionalism being reinforced by a loss of population. Reher and Iriso-Napal also find

a negative correlation (-.503) between the leftist vote in 1936 and rural fertility levels,

9 For a further analysis of the link with infant mortality, see F. Gil-Alonso 2010.

Factor score

-2.57 - -0.70

-0.70 - 0.05

0.05 - 0.35

0.35 - 0.74

0.74 - 3.14



which is equally reflected in our findings for the marital fertility factor for provinces as

a whole.

We also explored the effect of primogeniture as a characteristic of the inheritance

system. The protection of property for the oldest son could indeed explain why

Catalonia was so far ahead with its fertility decline. However, the map produced by

Zapatero Molinero (1977: 19) shows that the primogeniture zone stretched westward to

include Huesca, Navarra, and the Basque country as well. Furthermore an Asturian-

Galician band also exists with a dominance of primogeniture or testamentary

preference. Some of these regions (e.g., Huesca) were equally at the forefront of

fertility control, but others were not at all (e.g., Navarra). Moreover, the three provinces

with fast fertility decline in the Comunidad Valenciana all had equal division as the

dominant inheritance principle.

The conclusion from this exploration is that there is a modest link of the Spanish

marital fertility transition to industrialization, urbanization, and secularization, but that

major regional particularities prevail, even if further controls are introduced for

inheritance or land tenure systems.

4.3.2. The FDT nuptiality system

There are major differences between Belgium and Spain with respect to the historical

patterns and changes in their respective nuptiality systems. Belgium fell entirely within

the broader western European zone with the dominance of the Malthusian system of

late and non-universal marriages during the 18th

and 19th

centuries. This was not the

case for Spain, where the Malthusian system only prevailed in the Atlantic provinces

(see Map 14 and Reher 1991). In these areas the Hutterite fertility weighted Princeton

index Im of proportions of women married at ages 15 to 49 are in the range of .440 to

.480 in 1887 (Reher et al. 1993: 227). At the other end of the spectrum, Im values in

1887 are as high as .680 in several Spanish provinces (e.g., Avila, Huesca, Segovia)

(ibidem), which is indicative of much earlier marriage. By contrast, the Im levels for

Belgian arrondissements in 1880 were systematically lower than those in Spain, and

only in the .350 to .500 range, reflecting the impact of stronger Malthusian control.

A direct consequence of this distinction at the onset is that the proportion of

married women rose during the FDT in Belgium as the Malthusian system gradually

receded under the impact of the growth of wage labor in industrial sectors (Lesthaeghe

1977). By contrast, in Spain the proportions married experienced a modest decline. In

terms of the Princeton index Im, there was a decrease from .575 in 1887 for Spain as a

whole to .504 in 1930, and in terms of mean ages at first marriage for women there was



a small increase from 24.4 in 1900 to 24.7 in 1930 (Muñoz-Perez and Recaño-Valverde

2011).

The spatial nuptiality pattern is shown in Map 14 and captures the Im values for

1887, 1900, and 1930. The best correlates of this map deal with the land tenure system.

The minifundio indicator, i.e. percent of plots of less than one hectare in 1962 (INE,

1964) and the latifundio indicator of plots of over 100 hectares have correlation

coefficients of -.594 and .395, respectively. This reflects the correspondence between

late marriage and small-scale subsistence farming in Galicia and Asturias, and much

earlier marriage among the rural proletariat in central and southern regions with large-

scale agricultural enterprises. The correlation is not stronger because other areas existed

with small scale farming along the Mediterranean coast, but these did not necessarily

have the late nuptiality pattern (e.g. Alicante, Murcia).

Map 14: FDT - earlier female age at marriage, high proportion of women

married 15-49, 1887-1930

But the land tenure system does not tell the whole story. At least as important, but

connected to population pressure in densely populated rural areas, is the imbalance in

the sex ratios at marriageable ages, say 20-29. This ratio of men 20-29 to women 20-29

Factor score

-2.61 - -0.57

-0.57 - 0.14

0.14 - 0.52

0.52 - 0.82

0.82 - 1.09



for all the Atlantic provinces, from Gipuzkoa to all of Galicia and also including Leon

and Zamora, shows a large deficit of men, thereby reducing the marriage possibilities of

women. Hence selective outmigration of men depressed the values of Im considerably.

This feature, already described for 1887 (Reher et al. 1993), continued to exist well into

the 20th

century.

The Belgian values of Im in the arrondissements and towns show very similar

effects of selective migration by gender and marital status (Lesthaeghe 1977: 80-87).

Densely settled rural areas with cottage industries and small scale agriculture similarly

lose young men and have the lowest Im values. The older urban areas, which offer

employment opportunities for single women in textile industries, education, and

domestic service, have higher overall female employment rates and a shortage of single

men, which reduces the Im values as well. On the other hand, the new industrial towns

with heavy industries and growing suburbs had ample male-oriented employment

opportunities, a surplus of single men, and hence a favorable market for women, and

much higher Im values as a result (often above .550) (Ibidem: 84-85).

To sum up, in Belgium the gradual abandonment of the late Malthusian marriage

pattern was a major ingredient of the FDT and there is a steady drop in ages at marriage

until the mid-1960s. In Spain, the situation is much more stable until the 1930s, and the

decline in marital fertility is not really accompanied by a nuptiality transition as in

northern or western Europe. Instead, nuptiality patterns partially continued to reflect the

strong differences in land tenure systems, with later marriage in areas with peasant

agriculture, and early marriage in areas with latifundio and a rural proletariat.

In both countries there were major additional “marriage market” effects caused by

selective migration by gender and marital status. Whenever migration creates a surplus

of young women, Im values are depressed, and the opposite holds when a surplus of

young men is produced. There is a strong push for male outmigration in areas with

small-scale peasant agriculture and high population densities, leading to low Im values,

and hence the “land tenure – migration” link is of extra importance.

4.3.3 The “illegitimacy” dimension of the FDT

Out-of-wedlock fertility levels in Spain and Belgium were low to very low. For Spain

as a whole the Princeton index of illegitimacy Ih amounted to .041 in 1887 and it

declined to barely .038 in 1930 (Coale and Treadway 1986: 145). The zones with the

highest levels of illegitimacy, i.e. with Ih values between .060 and .085, are all in the

south. The other areas with higher illegitimacy (Ih values between .050 and.065) are

located in Galicia. Finally there are also such values in Madrid, and more surprisingly

in the province of Valladolid.



The major correlates of higher illegitimacy are the percentages urban in 1900 and

1950, with correlation coefficients of .468 and .412 respectively. This is essentially the

same finding as in the case of Belgian illegitimacy. It is also possible that the higher

illegitimacy in southern Spanish provinces reflects the remnants of unmarried

cohabitation, but the overall correlation with, for instance, the presence of latifundio is

not materializing because of other areas with similarly higher illegitimacy levels,

reflecting other features such as urbanization.

Map 15: FDT - high out-of-wedlock fertility, 1887-1960

4.3.4 The postponement dimension of the SDT

As indicated in Table 4, the SDT postponement dimension is essentially capturing the

proportions of women who are still not in a union at ages 25-29, the proportions of

women still childless in the age groups 25-29 and 30-34, and the fertility postponement

index, which is the ratio of the sum of the age-specific fertility rates above age 30 over

that in the age bracket 20-29. All these indicators pertain to the year 2001. Another

good correlate of this factor was lower total fertility as measured by the TFR in 1995.

Factor score

-1.64 - -0.78

-0.78 - -0.24

-0.24 - 0.11

0.11 - 0.95

0.95 - 2.95



The geography of the SDT postponement dimension is displayed in Map 16. The

slowest transitions into marriage and especially to parenthood are found in a large

contiguous set of provinces. The core of it is made up of Navarra, the Basque region,

Rioja, Soria, Burgos, Palencia, Valladolid, and Madrid. Around this core is another

circle of provinces with strong SDT postponement, so that the whole of Castile and

Leon is covered, together with northern and central Aragon and the province of

Guadalajara. The outlier here is Barcelona, which is the only province with stronger

postponement outside that contiguous area.

The postponement is less pronounced in the rest of Spain, and least of all in

Extremadura, the adjacent part of Castile & Leon, Andalusia (minus Seville), and

Murcia. At the other end of the country Pontevedra in Galicia also falls within the

lowest quintile of this postponement dimension.

Map 16: SDT - postponement of marriage, parenthood, and overall fertility,

1995-2001 (quintiles)

The most striking correlates of the SDT postponement factor all deal with literacy

and education: the total population literacy rate in 1887 (r = .814), the same rate for the

population aged 10+ in 1910 (.798), the percentage of women 10+ still incapable of

Factor score

-1.91 - -0.86

-0.86 - -0.41

-0.41 - 0.16

0.16 - 0.97

0.97 - 2.03



reading and writing as late as 1986 (-.806), and the much more recent percentage of

women 16+ in 2001 with more than complete secondary education (.797). Hence there

is a striking continuity over time of the impact of the education factor from illiteracy in

1887 to tertiary education in 2001. In order to bring this out even more clearly, Figure 1

shows the scattergram illustrating the location of the provinces in this spatial correlation

between the SDT dimension and population literacy measured as early as in the 1887

census.

Figure 2: Scattergram for the correlation (r = +0.814) between the SDT

postponement dimension (1995-2001) and the percentage literate in

the total population in 1887

The other major correlates of the SDT postponement dimension are equally

correlates of the literacy/education variables, and, as outlined before, they will also

reflect the connection between education and a stronger historical attachment to

Catholicism. Hence, the SDT postponement factor also correlates with the number of

persons per clergyman in 1887 (-.693) and with the index of marriages in Lent in 1900-

1901 (-.415). Better education for women is also reflected in higher employment rates,

and hence these rates for women 25-29 and 30-34 in 2001 are good predictors of SDT

postponement (r = .685 and .672, respectively). Finally, the daily press circulation in



1995, another indicator of the population educational level, is also a fair correlate (.511)

of starting marriage and parenthood later.

4.3.5 The “non-conformism” dimension of the SDT

The “non-conformism” dimension of the SDT is captured by four indicators for the

period 1995-2002: the percentage of all births outside marriage, the percentage of

cohabiting unions, the abortion rate for women 15-44, and the percentage of children

living in single parent households (see table 4). Other strong demographic correlates of

this dimension were higher divorce+separation rates for women 15-44 in 1991 and

2001 (r = .689 and .651, respectively).10

Non-conformism in current family formation is found in several zones, as shown

in Map 17.

The first zone stretches along the entire Mediterranean coast, from the whole of

Catalonia to Cadiz, and including the Baleares Islands. The second one comprises the

Basque provinces, Navarra and Rioja. The third area is made up of Madrid and adjacent

Guadalajara, and the fourth one of the Canary Islands.

The most significant correlates of this SDT non-conformity factor essentially

capture the degree of urbanity and concomitant employment in the service sector, in

tandem with the “new secularization” dimension. In more detail, the urbanity dimension

shows up in correlations with the percentage employed in the service sector in 1995 (r =

.755), in agriculture in 1995 (-.678), the percentage of women 16+ with secondary

education in 2001 (.625) and their activity rate in 2002 (.608). The effect of the “new

secularization arc” is reflected in the correlations between the SDT non-conformity

factor and the percentages with only civil marriages in 2001 (r = .529) and 2009 (r =

.519). But there are also echoes of the past. The urban percentages in 1900 and 1950

have a correlation with the SDT non-conformity factor of .463 and .458, and the index

of marriages in Lent of 1900-01 a correlation of .437.

10 See also Cabré-Pla and Menacho-Montes 2007.



Map 17: SDT – non-conformist family formation (cohabitation, out-of

wedlock fertility, single parent household, abortion), 1995-2002

5. A common multivariate analysis strategy for both countries

So far we have only discussed patterns of zero order correlations between the various

demographic dimensions and their predictors in Belgium and Spain, but a directly

comparable multivariate analysis is needed to help us further. One of the problems of

earlier detailed work presented by Reher and Iriso-Napal (1987, 1989) is the presence

of strong multicollinearity. This emerges when predictors are fairly strongly correlated

between themselves, and particularly so when such predictor intercorrelations are

stronger than those with the dependent variable. Multicollinearity causes standardized

beta-coefficients in multiple regressions to be substantially different from the zero-order

correlations, and also causes them to have enlarged standard errors. In other words,

estimates of net effects become unstable. In the instance of Spain this is a special worry,

because the correlations between the presumed structural or cultural determinants and

the FDT demographic dimensions are not very strong to start with. Another point is that

Factor score

-1.93 - -0.84

-0.84 - -0.12

-0.12 - 0.16

0.16 - 0.79

0.79 - 3.15



Reher and Iriso-Napal split their provincial data set into separate urban and rural

subsets, so that no direct comparison with their findings can be made.

In this application we shall stick to a single data set for each of the countries and

enter urbanization as a variable, rather than using it for splitting up the data. But, more

importantly, we shall opt for a different analysis strategy altogether, with the aim of

eliminating the multicollinearity problem in the estimation of the net effects of

determinants.

In the analysis so far, a clearer picture of the FDT and SDT dimensions could be

obtained by subjecting the various indicators to a Principal Components Analysis. By

imposing an orthogonal design of factor extraction, uncorrelated dimensions could be

obtained which prove to have very clear interpretations. Now, exactly the same strategy

will be followed for the independent variables. The extracted factors or dimensions will

also be uncorrelated between themselves, and the standardized regression coefficients

(betas) of such uncorrelated predictors will be identical to the original zero order

correlation coefficients. On the other hand, a factor may capture two strongly correlated

features, such as industrialization and secularization or education and Catholicism. In

such instances the effects of these two cannot be estimated separately. What we get is

the effect of the synergy of the two together.11

5.1 The principal components of the independent variables sets in Belgium and

Spain

In the Belgian example 22 indicators of structural, social, and cultural determinants of

the FDT and SDT could be reduced to four main orthogonal dimensions while retaining

80% of the total variance of the indicators. In the Spanish example six orthogonal

components were needed to capture 78% of the total variance contributed by 27

indicators. The factors and their main indicators are presented in Tables 5 and 6 for the

two countries respectively. The values between parentheses are the factor loadings or

the zero order correlation coefficient between the factor and the indicator concerned.

11 There are generally very good historical reasons that explain such combinations of determinants, and these

can be made explicit through a more detailed narrative rather than through purely statistical causal modeling.



Table 5: The four orthogonal dimensions of the socio-economic and cultural

predictors in 42 Belgian arrondissements and the factor loadings of

the indicators (r)

Factor 1. Urbanity . Factor 2: Secularization and

pre-World War I subsistence

agriculture.

Factor 3: Literacy. Factor 4: Advancement

female education after World

War II

Percent population

urban 1890 (%urb) 0.734

Vote for secular

parties 1958 0.938

Literacy levels

population 15+ (lit

15+)1880

0.947

Percent women in

birth cohort 1931-35

with complete

secondary or higher

education

0.720

% urb 1900 0.786 Vote for secular

parties 1919 0.965 Lit 15+ 1890 0.942

Percent women in

birth cohort 1961-65

with tertiary education

0.940

% urb 1910 0.787 Non-attendance

Sunday Mass 1964 0.895 Lit 15+ 1900 0.926

Ever worked women

cohort 1931-35 0.472

% urb 1920 0.885 MLA-index 1881-1884 0.797 Lit 15+ 1910 0.933 Ever worked women

cohort 1961-65 0.572

% urb 1930 0.880

Percent male active

population in

agriculture (% male

agr.) 1890

-0.842

% urb 1947 0.884 % male agr. 1900 -0.792

% urb 1961 0.854 % male agr. 1910 -0.755

% urb 1970 0.887

Percent male active

population in agr. and

cottage industries

1900

-0.822

Percent male active

population in

agriculture and

cottage industries,

1900

-0.585

percent population

rural 1970 -0.642

Tables 7 and 8 show, for Belgium and Spain respectively, the results of OLS

multiple regression. There is zero multicollinearity since the predictor dimensions are

uncorrelated among themselves.



Table 6: The six orthogonal dimensions of the socio-economic and cultural

predictors in 50 Spanish provinces and the factor loadings of the

indicators (r )

Factor 1. Early industrialization. Factor 2: Urbanity and service sector. Factor 3: Land tenure – Minifundio vs

Latifundio

Percent male employment in

industry (% male ind.) 1900 0.643

Percent population urban (%

urb) 1900 0.852

Percent agricultural

enterprises less than 1ha

1986

0.859

% male ind. 1950 0. 585 % urb 1950 0.877 Promille agricultural

enterprises 100+ ha 1986 -0.679

Percent males in agriculture

(% male agr.) 1887 -0.412

Percent employment in

industry (% ind.) 1900 0.587

Average size agricultural

enterprises 1962 -0.659

Percent diputados right parties

elections 1933 -0.520 % ind. 1950 0.528

Net migration rate 1887 0.443

Percent employment in service

sector 1995 0.712

Percent employment in

agriculture 1995 -0.793

Percent vote left 1977 0.724

Factor 4. Historical literacy and

Catholicism.

Factor 5: Traditional left Factor 6: “New secularization” and

contemporary female employment

Percent population literate (%

lit.) 1887 0.829

Percent diputados left parties

1936 0.804

Percent civil marriages (% civil

marr.) 2001 0.900

% lit. 1900 0.753 Percent diputados right

parties 1931 -0.705 % civil marr. 2009 0.895

Percent women 15 still illiterate

1986 -0.666

Percent diputados left parties

1931 0.603

Female employment rate 30-

34 in 2001 0.903

Percent women 16+ with

tertiary education 2001 0.617 Percent vote left 1977 0.377

Female employment rate 25-

29 in 2001 0.750

Number of persons per

clergyman 1887 -0.634

Percent women16+ with

secondary educ. 2001 0.691

Percent diputados right parties

elections 1933 0.632

Daily press circulation index

1995 0.756

Index marriages Lent, 1900-01 -0.574 PNB per capita 1960 0.643

Ratio patronos/non-patronos in

agriculture 1920 0.535

Percent women 10+ still

illiterate 1986 -0.607



Table 7: The prediction of the three demographic dimensions in Belgian

arrondissements on the basis of the four independent variable

factors. Results from multiple regression: beta coefficients and

adjusted R squared values

N=42 FDT+SDT continuity: Fertility

control + non-conformism

family formation

FDT: Historical illegitimate

fertility (Ih)

SDT: postponement fertility.

Urbanity .181* .578*** .188*

Secularization and pre-WWI

subsist. agriculture .840*** .413*** -.121

Literacy .377*** -.437*** .167

Advancement postwar female

education .090 .167 .801***

Adj. R squared .873 .687 .691

(R) ( .941) (.847) (.849)

Note: *** p<.001 , **p<.01, *p<.05.

Table 8: The prediction of the five demographic dimensions in Spanish

provinces on the basis of the six independent variables factors.

Results from multiple regression: beta coefficients and adjusted R

squared values

N=50 FDT: high

marital fertility

Ig

FDT : earlier

marriage Im

FDT: higher

non-marital

fertility Ih

SDT:

postponement of

unions & fertility

SDT: “non-

conformism”

family formation

Early

industrialization -.313* -.097 -.235 -.023 -.177

Urbanity -.096 .154 .412*** .005 .571***

Minifundio (vs

latifundio) .071 -.585*** .096 -.312*** -.009

High historical

literacy +

Catholicism.

.077 -.058 .222 .694*** -.302***

Traditional Left -.439*** .007 .264* -.195** -.091

“New secular.” +

high contem.

female emplmt

na na na .452*** .525***

Adj. R squared

( R )

.233

(.558)

.309

(.615)

.279

(.594)

.797

(.907)

.695

(.856)

Note: *** p<.001 , **p<.01, *p<.05. na = not applicable (not a historical covariate).



The first major contrast between the countries is that the adjusted multiple

correlation coefficients squared (or the proportions of the variance of the dependent

dimension explained) are much higher for the spatial patterning of the FDT in Belgium

than in Spain. For the SDT, however, there is no major distinction anymore in this

respect, since the adjusted R-squared values for Spain have become much higher. In

other words, the spatial patterns of the SDT dimensions in Spain are much better

accounted for than those of the FDT.

In both countries industrialization and urbanization have, not surprisingly,

contributed to a faster historical marital fertility decline, to a more pronounced bulge in

illegitimacy during the FDT, and to the stronger emergence of the “non-conformism”

dimension of the SDT. Note, however, that the impact of the urbanity dimension on the

Belgian first factor is not large. This is mainly due to the fact that the control of marital

fertility started in several Walloon rural areas even before spreading to the industrial

and urban areas.

Just as important as these structural aspects was the dimension of secularization. In

Belgium this was by far the strongest predictor of both behavioral “innovations”, i.e.

adoption of deliberate fertility control (FDT) and adoption of non-conventional family

formation patterns (SDT). In Spain, the old secularization pattern associated with the

political Left was equally instrumental in speeding up the fertility transition, whereas

the “new secularization” pattern accounts more strongly for the “non-conformity” part

of the SDT. In both countries historical illegitimate fertility levels were also higher in

more secularized areas.

The impact of historical literacy levels in Belgium is noticeable when it comes to

speeding up fertility control and reducing illegitimacy during the FDT, but it vanishes

for the SDT postponement dimension. By contrast, the Spanish concordance of

Catholicism with better education is a very strong one throughout the 20th

century, and

it currently accounts for higher proportions of women with post-secondary education.

Through this long historical chain, this factor has become the best predictor of the

fertility postponement dimension of the SDT. In Belgium there was no such clear

connection between religion and more advanced education, and the postponement

dimension of the SDT is mainly explained by just female higher education and higher

employment rates.

Finally, a specific Spanish feature is that the spatial differences in nuptiality are

partially accounted for by the minifundio-latifundio contrast. Regions with high rural

population densities and a preponderance of small-scale agriculture had higher

proportions of single women. This was further reinforced by high male outmigration, a

sex imbalance, and concomitant unfavorable marriage markets for women. In Belgium

the gradual weakening of the Malthusian late marriage pattern was mainly associated

with the growth of urban and industrial wage sectors, but sex imbalances played a



similar role as in Spain, i.e. more unfavorable marriage conditions for the

overrepresented gender.

5.2 Conclusions: Back to the RWA paradigm

The RWA-model of demographic change essentially implies that new forms of

behavior will mainly develop spatial patterns that will mirror the “Willingness”

condition if such forms of behavior run counter to pre-existing moral or ethical codes of

conduct. The practice of deliberate contraception (FDT), and then much later also the

replacement of legal marriage by cohabitation and concomitant parenthood within such

cohabiting unions (SDT), were two such demographic innovations that challenged

moral conventions and religious doctrine. Hence the normal outcome should be that

these two features of FDT and SDT respectively should mirror the map of

secularization. And since the latter had clearly crystallized by 1880 in both Spain and

Belgium, then it is perfectly logical that the maps of the onset of fertility control (FDT)

and of the rise of cohabitation (SDT) in these countries should be affected by the

subsequent evolution of secularization. The empirical investigation presented here for

the two countries indicates that this has indeed been the case, and quite strongly so. But

there is still a difference between the two countries. In Belgium the regional pattern of

secularization has remained very stable ever since the end of the 19th

century, and as a

result the “non-conformist” part of the SDT follows in the footsteps of the historical

marital fertility transition. In Spain the secularization map changed after the Franco

period with the growth of the “peripheral arc” as a continuous zone of high

secularization and the disappearance from it of Andalusia. Consequently, the old

secularization map predicts the historical marital fertility decline, whereas the new

secularization map predicts the “non-conformist” dimension of the SDT. To sum up, in

both countries the correlation with secularization remains through time, but in Belgium

this occurs within the context of stable differentials and in Spain within the context of

changing spatial patterns of secularization.

The fertility postponement ingredient of the SDT provides an illustration of a very

different outcome. Waiting to have children till later, i.e. until education is completed or

until a sustainable standard of living is reached, does not run counter to any moral or

ethical prescripts. In fact quite the opposite is true (and has been true in the West even

before Malthus defined his “prudent” marriage rules). Hence the limiting condition for

this aspect of the SDT should not be related to the “Willingness” factor (moral code) in

the RWA model. Rather, postponement of parenthood should reflect socio-economic

conditions, and hence be influenced mainly by structural economic constraints, captured

by the “Readiness” condition. This seems to be exactly the case with the spatial pattern



of the SDT fertility postponement in both countries, as shown by the strong spatial

correlation between parenthood postponement and female higher education and labor

force participation. In other words, the outcome is no longer reflecting the leads and

lags with respect to a moral and religious dimension, but much more reproducing the

map of structural socio-economic conditions. The outcome is that fertility

postponement is a newer feature, not dependent on moral legitimation, and therefore

spatially uncorrelated with the spread of cohabitation in both countries examined here.

The long-term effect of female literacy, and later on of female higher education, on

fertility postponement during the SDT is a specific additional Spanish feature.

The “Ability” condition has not been considered in the present analysis. But this

does not mean that it is unimportant or unrelated to the dimensions studied here. For

one, the availability of contraceptive methods, access to safe abortion, or the possibility

of making contracts between cohabiting partners are three issues that are all ultimately

settled at the political level. In both Belgium and Spain Catholic conservatism hindered

the sale of contraceptives for decades and dramatically opposed abortion. By contrast,

secular parties opened the way for free access to these methods and furthermore made

legal provisions to accommodate cohabitation and even same sex marriage. Hence, the

secularization dimension pops up in the “Ability” condition as well, and very forcefully

so in the two countries considered here.

Finally, Spain and Belgium share another striking feature: language borders are

capable of demarcating separate areas of diffusion of new forms of behavior. For

Belgium this is a bipolar division, and for Spain a multipolar one. Moreover, these

demarcations are best visible with regard to both the fertility decline of the FDT and the

“non-conformity” dimension of the SDT. Very often the same divisions emerge with

respect to secularization as well. This configuration is very visible in Belgium with the

Walloon-Flemish language border showing up each time these dimensions are

considered. Wallonia is simply systematically ahead with respect to fertility control

(FDT), rise of cohabitation, and parenthood among cohabitants (SDT), and this

corresponds to a long history of earlier and stronger secularization. In Spain a similar

role is played by the Catalan-Valencian linguistic zone. All the provinces in this area

(Catalonia, Comunidad Valenciana, Baleares Islands) tend to move together as an

“avant-garde” with respect to marital fertility decline and the SDT dimension of non-

conformist family formation. The entire zone is equally always present in both the old

secularization and new secularization maps. The Andalusian, Galician, and Basque

areas equally have their particularisms that set them apart from the Spanish core, made

up of the two Castiles and Extremadura, but the division along the demographic

dimensions is not as systematic as in the case of the Catalan-Valencian area.

Nevertheless, it comes as no surprise that, back in 1961, William Leasure’s professor of



Spanish in Princeton thought that Leasure had drawn up a map of Spanish languages,

when in fact he was presenting a map of the Spanish fertility transition.

In a nutshell, the main conclusions are:

(i) Regional differences with respect to the manifest control of fertility within

marriage (FDT) and “non-conformist” patterns of household formation (SDT) tend to

mirror the maps of secularization, and are hence especially conditioned by the

“Willingness” factor (W). By contrast the fertility postponement aspect of the SDT

reflects spatial patterns of female education and employment, and is consequently

essentially conditioned by the “Readiness” factor (R). These generalizations hold in

both countries.

(ii) However, in Belgium spatial continuity from FDT to SDT is linked to the

country’s stable pattern of secularization, whereas in Spain the conditioning element of

long-term continuity has been first female literacy and subsequently the growth of

female education.

(iii) The two sub-dimensions of the SDT, i.e. “non-conformist” family formation

and fertility postponement, are not nearly as interconnected as originally assumed by

the SDT theory. This not only emerges in Belgium and Spain but was also clearly in

evidence for the counties of the USA (Lesthaeghe and Neidert 2006) and shows up

again in the regions of Latin America (Esteve et al. 2012). The reason for this seems to

be the connection of the two demographic sub-dimensions to the different conditioning

factors mentioned above (W versus R).

6. Acknowledgements

Support for this research came from the WorldFam project (ERC-2009-StG-240978)

and the Center for Demographic Studies, Barcelona. The authors would also like to

thank Profs. D. Reher, A. Cabré and J. Recaño-Valverde for their preliminary

comments. These acknowledgements should further be extended to an anonymous

reviewer who contributed very useful comments and suggestions.



References

Bocquet-Appel, J.-P. and Jakobi, L. (1996). Barriers for the spatial diffusion of the

demographic transition in Europe. In: Bocquet-Appel, J.-P., Courgeau, D.D., and

Pumain, D. (eds.). Spatial analysis of biodemographic data. Paris & London:

Eurotext & John Libbey: 117-129.

Botti, A. (1992). Cielo y dinero. El “nacionalcatolicismo” en España (1881-1975).

Madrid: Alianza.

Cabré-Pla, A. and Menacho-Montes, T. (2007). Indicatores sobre la constitucion

familiar en España y sus provincias, 1991-2001. In: Cabré-Pla, A. (ed.). La

constitucion familiar en España. Bilbao: Fondacion BBVA: 297-340.

Coale, A.J. (1965). Factors associated with the development of low fertility: A historic

summary. New York: World Population Conference, United Nations. (Vol. 2:

207ff).

Coale, A.J. (1973). The demographic transition reconsidered. Liège: International

Population Conference, International Union for the Scientific Study of

Population (IUSSP). 1: 53-72.

Coale, A.J. and Treadway, R. (1986). Summary of the changing distribution of overall

fertility, marital fertility and the proportion married in the provinces of Europe.

In: The decline of fertility in Europe. Princeton N.J.: Princeton University Press.

Deboosere, P., Lesthaeghe, R., Surkyn, J., Willaert, D., Boulanger, P.-M., Lambert, A.,

and Lohlé-Tart, L. (2009). Huishoudens en gezinnen in België. Monografieën

Sociaal-Economische Enquête 2001, Nr 4. Brussels: Ministry of Economic

Affairs, Directie Statistiek & Economische Informatie (ADSEI), and Federale

Diensten voor Wetenschappelijke, Technische, en Culturele Aangelegenheden

(FDWTC). http://www.vub.ac.be/SOCO/demo/projects/monohouseholds.htm.

Esteve, A., Lesthaeghe, R. Lopez-Gay, A. (2012). The Latin-American cohabitation

boom, 1970-2007. Population and Development Review 38(1): 55-81.

doi:10.1111/j.1728-4457.2012.00472.x.

http://dx.doi.org/10.1111%2Fj.1728-4457.2012.00472.x



Gadeyne, S. , Neels, K., and De Wachter, D. (forthcoming): Nuptialiteit en

Vruchtbaarheid. Monografieën Sociaal-Economische Enquête 2001, Nr 5.,

Brussels: Ministry of Economic Affairs, Directie Statistiek & Economische

Informatie (ADSEI), and Federale Diensten voor Wetenschappelijke,

Technische, en Culturele Aangelegenheden (FDWTC).

Galloway, P.R. (2009). Fertility decline and social democracy in Prussia 1875 to 1910.

http://www.patrickrgalloway.com/galloway_2009_fertility_decline_and_social_

democracy_in_prussia_1875_to_1910.pdf

Galloway, P.R., Lee, R.D., and Hammel, E.A. (1994). Fertility decline in Prussia, 1875

to 1910 – A pooled cross-section time series analysis. Population Studies 48(1):

135-158. doi:10.1080/0032472031000147516.

Instituto Nacional de Estadistica (INE) (1964). Numero y superficie de las

explotaciones. Censo de 1962. Madrid.

Instituto Nacional de Estadistica (INE) (2005). Indicadores sociales de España, 2004.

Madrid.

Iriso-Napal, P.L. and Reher, D.S. (1987). La fecundidad y sus determinantes en España,

1887-1920. Un ensayo de interpretacion. Revista Española de Investigaciones

Sociologicas 39(87): 45-118. doi:10.2307/40183293.

Leasure, J.W. (1962). Factors involved in the decline of fertility in Spain, 1900-1950.

[Ph.D. Thesis]. Princeton: Dept of Economics, Princeton University. University

Microfilms, Ann Arbor, nr. 63-532.

Leasure, J.W. (1968) Factors involved in the decline of fertility in Spain, 1900-1950.

Population Studies 16(3): 271-285.

Lesthaeghe, R. (1977) The decline of Belgian fertility, 1800-1970. Princeton N.J.:

Princeton University Press.

Lesthaeghe, R. and Wilson, C. (1986). Modes of production, secularization and the pace

of the fertility decline in western Europe, 1870-1930. In: Coale, A.J. and

Watkins, S.C. (eds.). The decline of fertility in Europe. Princeton N.J.: Princeton

University Press: 261-292.

Lesthaeghe, R. (1991). Moral control, secularization and reproduction in Belgium

(1600-1900). In: Historiens et populations. Louvain-la-Neuve: Liber Amicorum

Etienne Hélin, éditions Académia: 259-279.

http://dx.doi.org/10.1080%2F0032472031000147516

http://dx.doi.org/10.2307/40183293



Lesthaeghe, R. (1998). On theory development and applications to the study of family

formation. Population and Development Review 24(1): 1-14. doi:10.2307/

2808120.

Lesthaeghe, R. and Neels, V. (2002). From a first to a second demographic transition –

An interpretationof the spatial continuity of demographic innovation in France,

Belgium and Switzerland. European Journal of Population 18: 325-360.

doi:10.1023/A:1021125800070.

Lesthaeghe, R. and van de Kaa, D.J. (1986). Twee demografische transities? In: Groei

en krimp. Mens en Maatschappij, annual book volume. Deventer: Van Loghum-

Slaterus: 9-24.

Lesthaeghe, R. and Neidert, L. (2006). The second demographic transition in the United

States – Exception or textbook example? Population and Development Review

32(4): 669-698. doi:10.1111/j.1728-4457.2006.00146.x.

Lesthaeghe, R. and Surkyn, J. (2006) When history moves on – Foundations and

diffusion of a second demographic transition. In: Jayakodi, R., Thornton, A., and

Axinn, V. (eds.). International family change: Ideational perspectives. New

York: Lawrence Erlbaum and Associates, Taylor & Francis Group: 81-118.

Lesthaeghe, R. and Vanderhoeft, C. (1999). Une conceptualization des transitions vers

des nouvelles formes de comportements. In: Tabutin, D. et al. (eds.). Théories,

paradigmes et courants explicatifs en démographie. Chaire Quételet 1997,

éditions. Louvain-la-Neuve: Académia-Bruylant: 279-306.

Lijphart, A. (1968). The politics of accommodation – Pluralism and democracy in the

Netherlands. Berkeley: University of California Press.

Linz, J.J. and Montero, J.R. (1999). The party systems of Spain: Old cleavages and new

challenges. Madrid: Center for Advanced Study in the Social Sciences, Juan

March Institut (Estudio/Working paper 1999/138).

Livi-Bacci, M. (1968). Fertility and nuptiality changes in Spain from the late 18th

to the

early 20th

Century. Population Studies 22(2): 63-102 and 211-234. doi:10.1080/

00324728.1968.10405536.

Livi-Bacci, M. (1971). A century of Portuguese fertility. Princeton N.J.: Princeton

University Press.

Livi-Bacci, M. (1977). A history of Italian fertility during the last two centuries.

Princeton N.J.: Princeton University Press.

http://dx.doi.org/10.2307%2F2808120

http://dx.doi.org/10.2307%2F2808120

http://dx.doi.org/10.1023%2FA%3A1021125800070

http://dx.doi.org/10.1111%2Fj.1728-4457.2006.00146.x

http://dx.doi.org/10.1080/00324728.1968.10405536

http://dx.doi.org/10.1080/00324728.1968.10405536



Muñoz-Perez, F. and Recaño-Valverde, J. (2011). A century of nuptiality in Spain,

1900-2007. European Journal of Population 27(4): 487-515. doi:10.1007/s106

80-011-9234-1.

Neels, K. (2006). Reproductive strategies in Belgian fertility, 1960-1990. Brussels:

Vlaamse Gemeenschap (NIDI-CBGS Publications nr 38).

Ragin, C. (1987). The comparative method – Moving beyond qualitative and

quantitative strategies. Berkeley: University of California Press.

Reher, D.S. (1991). Marriage patterns in Spain, 1887-1930. Journal of Family History

16(1): 7-30.

Reher, D.S. and Iriso-Napal, P.L. (1989): Marital fertility and its determinants in rural

and urban Spain, 1887-1930. Population Studies 43(3): 405-427. doi:10.1080/

0032472031000144216.

Reher, D.S., Nieves Pombo, M., and Nogueras, B. (1993). España a la luz del censo de

1887. Madrid: Instituto Nacional de Estadistica (INE).

Requeña, M. (2005) The secularization of Spanish society: Change in religious practice.

South European Society and Politics 10(3): 369-390. doi:10.1080/1360

8740500281902.

Sewell Jr., W.H. (2005): Logics of history – Social theory and social transformation.

Chicago IL.: University of Chicago Press.

Vilanova-Ribas, M. and Moreno-Julia, V. (1992). Atlas de la evolucion del

anafabetismo en España de 1887-1981. Madrid: CIDE, Centro de Publicaciones

del Ministerio de Educacion y Ciencia.

Zapatero Molinero, S. (1977). Sistema de herencia en la explotacion agraria. Zaragoza:

Ministerio de Agricultura, Instituto Nacional de Investigaciones Agrarias y

Centro Regional de Investigacion y Desarollo Agrario del Ebro.

http://dx.doi.org/10.1007/s10680-011-9234-1

http://dx.doi.org/10.1007/s10680-011-9234-1

http://dx.doi.org/10.1080/0032472031000144216

http://dx.doi.org/10.1080/0032472031000144216

http://dx.doi.org/10.1080/13608740500281902

http://dx.doi.org/10.1080/13608740500281902



Appendices

Map A1: Linguistic and administrative map of Belgium (provinces,

arrondissements)



Map A2: Linguistic map of Spain

Source: García Mouton, P. 1994 : Lenguas y dialectos en España. Editorial Arco Libros, Madrid.

CATALAN

ARAGONESE

BASQUE

ASTURIAN

GALICIAN

CASTILIAN

OCCITAN

Aragonese

Churro

ExtremeñoManchego

Andaluz

Murciano

Canario

CASTILIAN (Northern Variety)

CASTILIAN (Southern Variety)

NAME OF THE LANGUAGE

Name of the dialect



Map A3: Administrative map of Spain (Comunidades Autónomas and

provinces)

Notes: (1) Andalucía: Almería; Cádiz; Córdoba; Granada; Huelva; Jaén; Málaga; Sevilla. (2) Aragón: Huesca; Teruel; Zaragoza.

(3) Asturias, Principado de. (4) Balears, Illes. (5) Canarias: Las Palmas; Santa Cruz de Tenerife. (6) Cantabria.(7) Castilla y

León: Ávila; Burgos; León; Palencia; Salamanca; Segovia; Soria; Valladolid; Zamora. (8) Castilla- La Mancha: Albacete;

Ciudad Real; Cuenca; Guadalajara; Toledo. (9) Cataluña: Barcelona; Girona; Lleida; Tarragona. (10) Comunitat Valenciana:

Alicante/Alacant; Castellón/Castelló; Valencia/València. (11) Extremadura: Badajoz; Cáceres. (12) Galicia: A Coruña; Lugo;

Ourense; Pontevedra. (13) Madrid, Comunidad de. (14) Murcia, Región de. (15) Navarra, Comunidad Foral de. (16) País

Vasco: Araba/Álava; Bizkaia; Gipuzkoa. (17) Rioja, La.ga.

Zamora

Segovia

Madrid

Lugo

A Coruña

Asturias

Cantabria Vizcaya Guipúzcoa

Navarra

LeónAlava

Palencia

PontevedraHuesca Lleida

Zaragoza

Ourense

La Rioja Girona

Barcelona

Valladolid Soria

TarragonaSalamancaTeruel

GuadalajaraAvila

CastellónCuencaCáceres Toledo

Ciudad RealBadajoz Albacete

AlicanteCórdoba

Murcia

Jaén

Huelva

Sevilla Granada

AlmeríaMálagaCádiz

Valencia

Burgos

Baleares

Las Palmas

Santa Cruz de Tenerife

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)(11)

(12)

(13)

(14)

(15)

(16)

(17)

Spatial continuities and discontinuities in two successive ... · 2 The RWA model of innovation and...

Documents

Transcript of Spatial continuities and discontinuities in two successive ... · 2 The RWA model of innovation and...